JP2019140704A - Image prediction decoding device, and image prediction decoding method - Google Patents

Abstract

To enhance encoding efficiency.SOLUTION: According to one embodiment, an encoding target region in an image is divided into a plurality of prediction regions. A candidate for motion information used for creating a prediction signal for a target prediction region to be a next prediction region is selected from encoded motion information on a region adjacent to the target prediction region, on the basis of prediction information on an adjacent region adjacent to the target region, the number of encoded prediction regions in the target region, and encoded prediction information in the target region. According to the number of selected candidates for motion information, merge block information for instructing to create a prediction signal for the target prediction region using the selected candidate for the motion information and motion information detected by prediction information estimation means or any one of the merge block information and the detected motion information is encoded. The motion information used for creating the prediction signal for the target prediction region is stored in prediction information storage means.SELECTED DRAWING: Figure 1

Description

  One aspect of the present invention relates to an image predictive encoding device, an image predictive encoding method, and an image predictive encoding program. Another aspect of the present invention relates to an image predictive decoding apparatus, an image predictive decoding method, and an image predictive decoding program. In particular, these aspects include an image predictive encoding device, an image predictive encoding method, an image predictive encoding program, an image predictive decoding device, and an image predictive decoding method that generate a prediction signal of a target block using motion information of surrounding blocks. And an image predictive decoding program. Furthermore, another aspect of the present invention is a moving image encoding apparatus, a moving image encoding method, a moving image encoding program, a moving image decoding apparatus, and a moving image decoding that generate a prediction signal by motion compensation using a motion vector. The present invention relates to a method and a moving picture decoding program.

  In order to efficiently transmit and store still images and moving image data, compression coding technology is used. In the case of moving images, MPEG-1 to 4 and ITU (International Telecommunication Union) H.264. 261-H. H.264 is widely used.

  In these encoding methods, an encoding process or a decoding process is performed after an image to be encoded is divided into a plurality of blocks. In predictive coding within a screen, a predicted signal is generated using an adjacent previously reproduced image signal (reconstructed compressed image data) in the same screen as the target block, and then the predicted signal is The differential signal obtained by subtracting from the signal of the target block is encoded. In predictive coding between screens, an adjacent reproduced image signal in a screen different from the target block is referred to, motion correction is performed, a predicted signal is generated, and the predicted signal is subtracted from the signal of the target block. The difference signal obtained by this is encoded.

For example, H.M. H.264 intra-screen predictive encoding employs a method of generating a prediction signal by extrapolating already reproduced pixel values adjacent to a block to be encoded in a predetermined direction. FIG. 2 is a schematic diagram for explaining an intra-screen prediction method used for H.264. In FIG. 22A, a target block 802 is a block to be encoded, and a pixel group 801 composed of pixels P _{A to} P _M adjacent to the boundary of the target block 802 is an adjacent area, and a past This is an image signal that has already been reproduced in processing.

In the case illustrated in FIG. 22A, a prediction signal is generated by extending a pixel group 801 that is an adjacent pixel directly above the target block 802 downward. In the case shown in FIG. 22B, the prediction signal is generated by stretching the already reproduced pixels (P _{I to} P _L ) on the left of the target block 804 to the right. A specific method for generating a prediction signal is described in Patent Document 1, for example. Thus, the difference between each of the nine prediction signals generated by the method shown in FIGS. 22A to 22I and the pixel signal of the target block is taken, and the one with the smallest difference value is determined as the optimum prediction signal. To do. As described above, a prediction signal can be generated by extrapolating pixels. The above contents are described in Patent Document 1 below.

  In normal inter-screen predictive coding, a prediction signal is generated by searching for a signal similar to the pixel signal from a screen that has already been reproduced for a block to be coded. Then, a motion vector that is a spatial displacement amount between the target block and a region formed by the searched signal, and a residual signal between the pixel signal and the prediction signal of the target block are encoded. Such a method for searching for a motion vector for each block is called block matching.

  FIG. 21 is a schematic diagram for explaining the block matching process. Hereinafter, a procedure for generating a prediction signal will be described using the target block 702 on the encoding target screen 701 as an example. The screen 703 has already been reproduced, and the area 704 is an area in the same position as the target block 702. In the block matching, a search range 705 surrounding the region 704 is set, and a region 706 in which the absolute value error sum with the pixel signal of the target block 702 is minimum is detected from the pixel signal in this search range. The signal in the area 706 becomes a prediction signal, and the amount of displacement from the area 704 to the area 706 is detected as a motion vector 707. Also, there may be used a method of preparing a plurality of reference screens 703, selecting a reference screen for performing block matching for each target block, and detecting reference screen selection information. H. H.264 provides a plurality of prediction types having different block sizes for encoding motion vectors in order to cope with changes in local features of an image. H. H.264 prediction types are described in Patent Document 2, for example.

  In the compression encoding of moving image data, the encoding order of each screen (frame, field) may be arbitrary. For this reason, there are three types of coding order methods for inter-screen prediction in which a prediction signal is generated with reference to a reproduced screen. The first method is forward prediction in which a prediction signal is generated with reference to a past reproduced screen in the reproduction order, and the second method is a prediction signal with reference to a future reproduced screen in the reproduction order. This is a backward prediction to be generated, and the third method is bidirectional prediction in which both forward prediction and backward prediction are performed and two prediction signals are averaged. About the kind of prediction between screens, it describes in patent document 3, for example.

  In HEVC (High efficiency video coding), which has started standardization as a next-generation coded video coding scheme, the division type of a prediction block is divided into two rectangles shown in FIGS. 20B and 20C, and (D) of FIG. In addition to the tetragonal quadrant shown in FIG. 20, introduction of asymmetric division is also being considered as shown in FIGS. Furthermore, in the HEVC, when generating the prediction signal of the prediction block divided in this way, the motion information (motion vector, reference screen information, forward / backward / A method using an inter-screen prediction mode for identifying both directions has been studied. This prediction method is called block merging and has a feature that motion information can be encoded efficiently. (A) of FIG. 2 is a diagram schematically illustrating adjacent blocks when block merging is performed on a prediction block T1 generated by vertically dividing the encoded block 400. The prediction signal of the prediction block T1 is generated using either 1) motion information of the adjacent block A, 2) motion information of the adjacent block B, or 3) motion information detected by block matching. For example, when the encoder selects the motion information of the adjacent block A, the encoder first sets the merge identification information (merge_flag) indicating that the motion information of the adjacent block is used to “merge_flag = 1”. The merge identification information (merge_flag) is sent to the decoder. Second, the encoder sets merge block selection information (merge_flag_left) indicating that the adjacent block A is used from the adjacent block A and the adjacent block B to “merge_flag_left = 1”, and uses the merge block selection information (merge_flag_left). Send to the decoder. By receiving these two pieces of information, the decoder can identify that the prediction signal of the target prediction block should be generated using the motion information of the adjacent block A. Similarly, when the decoder receives “merge_flag = 1” and “merge_flag_left = 0” (selecting the adjacent block B), the decoder generates a prediction signal of the target prediction block using the motion information of the adjacent block B, and generates “merge_flag When “= 0” is received, it is further possible to identify that the motion information of the target prediction block should be restored by receiving motion information from the encoder. The block merging shown here is described in Non-Patent Document 1.

  Also, in inter-frame prediction in standards such as MPEG-1, 2 and MPEG-4, each picture is divided into a set of rectangular blocks that do not overlap each other, and a motion vector is associated with each block. The motion vector is obtained by a motion search for each block, and indicates the horizontal displacement and vertical displacement of the current block from the second block used for predicting the image signal of the current block.

  Patent Document 4 below describes a method for performing motion compensation prediction with higher accuracy when a motion boundary exists in a diagonal direction in a block. In this method, a block is further divided into non-rectangular small sections, and motion compensation prediction is performed for each small section.

  Patent Document 5 below describes a method of dividing a block into smaller rectangular small sections and performing motion compensation prediction for each small section. This method generates a predicted motion vector from the motion vector of a block that is in contact with the processing target sub-partition and is prior to the sub-partition in the processing order when encoding the motion vector of the processing target sub-partition. The difference between the motion vector and the predicted motion vector of the small section to be processed, that is, only the differential motion vector is encoded. In this method, when the processing target sub-partition is not in contact with the previous block in the processing order, the processing is performed from the motion vector of the other sub-partition in the processing order within the block including the processing target sub-partition. A predicted motion vector of the target small section is generated.

US Pat. No. 6,765,964 US Patent Publication No. 7003035 US Pat. No. 6,259,739 JP 2005-277968 A JP 2009-246972 A

Test Model under Consideration, Joint Collaborative Team on VideoCoding (JCT-VC) of ITU-T SG16 WP3 and ISO / IEC JTC1 / SC29 / WG11, 1st Meeting: Dresden, DE, 15-23 April, 2010, Document: JCTVC-A205

  In the above Non-Patent Document 1, selection of motion information candidates to be used in block merging of a plurality of prediction blocks obtained by dividing a target encoding block to be encoded is performed in the same manner regardless of the surrounding state of the prediction block. carry out. Therefore, for example, as illustrated in FIG. 2B, motion information candidates for generating a prediction signal of the prediction block T2 include motion information of the prediction block T1 included in the same encoded block. The prediction block division type including the prediction block T1 and the prediction block T2 is prepared on the assumption that the prediction signals of the two blocks are generated with different motion information. Therefore, it is not preferable that the motion information candidate of the prediction block T2 includes the motion information of the prediction block T1. That is, encoding may be inefficient.

  Therefore, according to some aspects of the present invention, motion information candidates used to generate a prediction signal of a target prediction block are encoded or decoded prediction information (motion of a target coding block and surrounding coding blocks). Image predictive coding device, image predictive coding method, image predictive coding program, image predictive decoding device, image predictive decoding, which suppresses generation of inefficient codes by selecting based on information and prediction block division type) It is an object to provide a method and an image predictive decoding program. That is, in these aspects, the present invention provides an image predictive encoding device, an image predictive encoding method, an image predictive encoding program, an image predictive decoding device, an image predictive decoding method, and an image capable of improving encoding efficiency. An object is to provide a predictive decoding program.

  Further, as described in Patent Document 4 or Patent Document 5, there is a method of performing motion compensation prediction for each small section obtained by dividing a block to be processed. In such motion compensated prediction, a predicted motion vector for a small block is generated based on the motion vector of the previous block in the processing order from the small block to be processed, and the motion vector of the small block and the predicted motion vector It is preferable from the viewpoint of the amount of code to encode only the difference motion vector between them.

  FIG. 23 is a diagram for explaining motion compensation prediction. As shown in FIG. 23, in the block P to be processed, there may be a small section SP1 in contact with one or more blocks CP preceding the block P and a small section SP2 not in contact with the block CP. The motion vector V2 of such a small section SP2 is encoded as it is without using the predicted motion vector in the method described in Patent Document 4. This method is equivalent to a method in which the predicted motion vector is a zero vector.

  On the other hand, in the method described in Patent Document 5, a predicted motion vector of the small section SP2 is generated from the motion vector V1 of the other small section SP1 in the block P and preceding the small section SP2 in the processing order. Is done. However, it is considered that the motion vector of the small section SP1 and the motion vector of the small section SP2 are inherently different from each other. Therefore, the method described in Patent Document 5 may not efficiently encode the motion vector of the small section SP2.

  In view of this, the present invention provides a moving image encoding apparatus, a moving image encoding method, a moving image encoding program, and a moving image that can improve the encoding efficiency in another aspect. It is an object of the present invention to provide a moving image decoding apparatus, a moving image decoding method, and a moving image decoding program corresponding to encoding.

One aspect of the present invention relates to image predictive decoding, and can be described as an image predictive decoding apparatus and method as follows.
The image predictive decoding device according to the present invention encodes prediction information that indicates a prediction method used to predict a signal of a target region to be decoded from compressed data of an image divided and encoded into a plurality of regions. Data analysis means for extracting encoded data and encoded data of the residual signal, and prediction information decoding means for restoring motion information from the encoded data of the prediction information, from the encoded data of the prediction information, A prediction block partition type indicating the number of prediction regions into which the target region is subdivided is restored, and the prediction block partition type is perpendicular to the first prediction region on the left side and the second prediction region on the right side. In the case of indicating a type to be divided into two, the encoded data of the prediction information is further decoded to generate a prediction signal of the first prediction region, and a screen of adjacent regions adjacent to the first prediction region is displayed. For forecasting First merge identification information indicating whether or not to use motion information candidates including decoded motion information attached to adjacent regions excluding the generated adjacent region is used, and the first merge identification information is the motion information When indicating that the information candidate is not used, the encoded data of the prediction information is further decoded to restore the first motion information used for generating the prediction signal of the first prediction region, and the first information When merge identification information indicates that the motion information candidate is to be used, the encoded data of the prediction information is further decoded and used for generating a prediction signal of the first prediction region from the motion information candidate The first selection information for identifying the motion information is restored, the first motion information is restored based on the first selection information, the encoded data of the prediction information is further decoded, and the second Generation of prediction signals for the prediction region The second merge identification information indicating whether or not to use the decoded motion information associated with the adjacent region adjacent to the second prediction region is restored, and the second merge identification information does not use the decoded motion information. The encoded data of the prediction information is further decoded to restore the second motion information used to generate the prediction signal of the second prediction region, and the second merge identification information has been decoded. A plurality of adjacent regions adjacent to the second prediction region by further decoding the encoded data of the prediction information when there is a plurality of candidates for the decoded motion information and indicating that motion information is used Among the motion information of a plurality of upper regions that do not include the first motion information and that are adjacent to the second prediction region above the second prediction region. Matches the movement information of Second selection information for identifying second motion information used for generating a prediction signal of the second prediction region is restored from the decoded motion information candidates that do not include the motion information to be reproduced, and the second selection is performed. The prediction information decoding means for restoring the second motion information based on the information, the storage means for saving the motion information included in the restored prediction information, the restored first motion information, and the second motion information. Prediction signal generating means for generating a prediction signal for each of the first prediction region and the second prediction region included in the target region based on the motion information of the target region, and the target signal from the encoded data of the residual signal Residual signal restoring means for restoring the reproduction residual signal of the area, and a recording for restoring the pixel signal of the target area based on the prediction signal and the reproduction residual signal and storing the pixel signal as an already reproduced signal Means.
An image predictive decoding method according to the present invention is an image predictive decoding method executed by an image predictive decoding device, and is a target to be decoded from compressed data of an image divided into a plurality of regions and encoded. A data analysis step of extracting encoded data of prediction information instructing a prediction method used for prediction of a signal in a region and encoded data of a residual signal, and prediction for restoring motion information from the encoded data of the prediction information An information decoding step, wherein a prediction block partition type indicating the number of prediction regions into which the target region is subdivided is restored from the encoded data of the prediction information, and the prediction block partition type includes the target region on the left side When the first prediction region and the second prediction region on the right side indicate a type of being divided into two vertically, the encoded data of the prediction information is further decoded to predict the first prediction region Whether to use a motion information candidate including decoded motion information associated with an adjacent area excluding an adjacent area generated by intra-screen prediction among adjacent areas adjacent to the first prediction area. The first merge identification information shown is restored, and when the first merge identification information indicates that the motion information candidate is not used, the encoded data of the prediction information is further decoded, and the first prediction When the first motion information used for generating the prediction signal of the region is restored, and the first merge identification information indicates that the motion information candidate is used, the encoded data of the prediction information is further decoded. First selection information for identifying first motion information used for generating a prediction signal of the first prediction region is restored from the motion information candidates, and the first motion is based on the first selection information. Information is restored and the prediction information A second code indicating whether or not the encoded data is further decoded and decoded motion information associated with an adjacent region adjacent to the second prediction region is used to generate a prediction signal of the second prediction region; When the merge identification information is restored and the second merge identification information indicates that the decoded motion information is not used, the encoded data of the prediction information is further decoded, and the prediction signal of the second prediction region When the second motion information used for generation is restored, the second merge identification information indicates that the decoded motion information is used, and there are a plurality of candidates for the decoded motion information, the prediction information The encoded data is further decoded, and the decoded motion information associated with a plurality of adjacent regions adjacent to the second prediction region does not include the first motion information and is located above the second prediction region. In the second prediction region The second used for generating a prediction signal of the second prediction region from the candidates of the decoded motion information not including the motion information that matches the first motion information among the motion information of the plurality of upper regions adjacent to the second region The prediction information decoding step of restoring the second selection information for identifying the motion information of the second and restoring the second motion information based on the second selection information, and the motion information included in the restored prediction information And a prediction step of each of the first prediction region and the second prediction region included in the target region based on the restored first motion information and the second motion information Based on the prediction signal generation step for generating the residual signal, the residual signal restoration step for restoring the reproduction residual signal of the target region from the encoded data of the residual signal, and the prediction signal and the reproduction residual signal Pixel of target area Restore the items, comprising a recording step of storing the pixel signal as the already reproduced signal.
The motion information may include information on motion vectors and reference screen numbers.
The following aspects of the present invention are also conceivable. The first aspect of the present invention relates to image predictive coding.

One aspect of the present invention relates to image predictive coding, and can be described as an image predictive coding apparatus, method, and program as follows.
The image predictive coding apparatus according to the present invention includes a region dividing unit that divides an input image into a plurality of regions, and a target region that is an encoding target divided by the region dividing unit is a first prediction region and a second prediction region. Subdividing into prediction regions, determining the number of prediction regions suitable for the target region and the prediction block partition type indicating the region shape, and generating a signal having a high correlation with the first and second prediction regions First and second motion information to be obtained from the respective prediction block generation type, generation of the prediction block division type, the first and second motion information, and the prediction signal of the first prediction region, First merge identification indicating whether or not to use a motion information candidate including at least one of decoded motion information and motion information indicating that a motion vector is 0 associated with an adjacent region adjacent to the first prediction region Information and the second prediction A second indicating whether or not to use the decoded motion information excluding the first motion information among the decoded motion information associated with an adjacent region adjacent to the second prediction region for generating a prediction signal of the region; Prediction information estimation means for obtaining prediction information including the merge identification information, prediction information encoding means for encoding prediction information associated with the target region, the first motion information, and the second motion information Based on the prediction signal generation means for generating the prediction signal of each of the first prediction region and the second prediction region, the prediction signal and the pixel signal of each of the first prediction region and the second prediction region A residual signal generating means for generating a residual signal based on the above, a residual signal encoding means for encoding the residual signal generated by the residual signal generating means, and encoded data of the residual signal Generate playback residual signal by decoding And the difference signal restoring means, the prediction signal and the based on the reproduced residual signal to generate a decompressed pixel signal of the target region, comprising a recording means for storing the reconstruction pixel signal as the already reproduced signal.
An image predictive encoding method according to the present invention is an image predictive encoding method executed by an image predictive encoding device, and is divided by an area dividing step for dividing an input image into a plurality of areas and the area dividing step. The target region to be encoded is subdivided into a first prediction region and a second prediction region, a prediction block division type indicating the number and region shape of prediction regions suitable for the target region is determined, First and second motion information for obtaining a signal having a high correlation with the first and second prediction regions from the already reproduced signal is predicted, respectively, and the predicted block division type, the first and second At least one of motion information and decoded motion information associated with an adjacent region adjacent to the first prediction region and motion information indicating that the motion vector is 0 for generating a prediction signal of the first prediction region Including motion Of the first merge identification information indicating whether or not to use the information candidate, and the decoded motion information accompanying the adjacent region adjacent to the second prediction region for generating the prediction signal of the second prediction region Prediction information estimation step for obtaining prediction information including second merge identification information indicating whether to use decoded motion information excluding the first motion information, and prediction information associated with the target region. A prediction information encoding step for encoding, and prediction signal generation for generating a prediction signal for each of the first prediction region and the second prediction region based on the first motion information and the second motion information A residual signal generation step for generating a residual signal based on the prediction signal and the pixel signal of each of the first prediction region and the second prediction region, and the residual signal generated by the residual signal generation step Encode the difference signal The target region based on the difference signal encoding step, the residual signal restoration step for generating a reproduction residual signal by decoding the encoded data of the residual signal, and the prediction signal and the reproduction residual signal Generating a restored pixel signal, and storing the restored pixel signal as the already reproduced signal.
An image predictive coding program according to the present invention includes a region dividing unit that divides an input image into a plurality of regions, and a target region that is an encoding target divided by the region dividing unit is a first prediction region. Subdividing into second prediction regions, determining the number of prediction regions suitable for the target region and the prediction block partition type indicating the region shape, and outputting a signal highly correlated with the first and second prediction regions First and second motion information to be obtained from the already reproduced signal is predicted, and the prediction block division type, the first and second motion information, and the prediction signal of the first prediction region are generated. First, indicating whether or not to use a motion information candidate including at least one of decoded motion information and motion information indicating that the motion vector is 0 associated with an adjacent region adjacent to the first prediction region Merge identification And the decoded motion information excluding the first motion information out of the decoded motion information associated with the adjacent region adjacent to the second prediction region is used to generate the prediction signal of the second prediction region Prediction information estimation means for obtaining prediction information including second merge identification information indicating whether or not to perform prediction information encoding means for encoding prediction information associated with the target area, and the first motion Prediction signal generating means for generating a prediction signal of each of the first prediction region and the second prediction region based on the information and the second motion information, and the first prediction region and the second prediction region A residual signal generating means for generating a residual signal based on each of the prediction signal and the pixel signal, a residual signal encoding means for encoding the residual signal generated by the residual signal generating means, By decoding the encoded data of the residual signal Residual signal restoring means for generating a raw residual signal, a restored pixel signal for the target area is generated based on the prediction signal and the reproduced residual signal, and the restored pixel signal is stored as the reproduced signal. It functions as a recording means.
One aspect of the present invention relates to image predictive decoding, and can be described as an image predictive decoding apparatus, method, and program as follows.
The image predictive decoding device according to the present invention encodes prediction information that indicates a prediction method used to predict a signal of a target region to be decoded from compressed data of an image divided and encoded into a plurality of regions. Data analysis means for extracting encoded data and encoded data of the residual signal, and prediction information decoding means for restoring motion information from the encoded data of the prediction information, from the encoded data of the prediction information, When the prediction block division type indicating the number of prediction regions into which the target region is subdivided is restored, and the prediction block division type indicates that the target region includes a first prediction region and a second prediction region Then, the encoded data of the prediction information is further decoded, and the decoded motion information and motion vector associated with the adjacent region adjacent to the first prediction region is 0 for generating the prediction signal of the first prediction region. The first merge identification information indicating whether or not to use a motion information candidate including at least one of the motion information indicating the presence is restored, and the first merge identification information indicates that the motion information candidate is not used. In this case, the encoded data of the prediction information is further decoded to restore the first motion information used for generating the prediction signal of the first prediction region, and the first merge identification information is the motion information candidate. Is used to further decode the encoded data of the prediction information to identify first motion information used for generating a prediction signal of the first prediction region from the motion information candidates Of the first prediction information, the first motion information is restored based on the first selection information, the encoded data of the prediction information is further decoded, and the prediction signal of the second prediction region is generated Adjacent to the second prediction region The second merge identification information indicating whether or not to use the decoded motion information associated with the adjacent region, and when the second merge identification information indicates that the decoded motion information is not used, The encoded data of the prediction information is further decoded to restore the second motion information used for generating the prediction signal of the second prediction region, and the second merge identification information uses the decoded motion information. If the encoded data of the prediction information is further decoded, the decoded motion information associated with a plurality of adjacent regions adjacent to the second prediction region is decoded to exclude the first motion information Second motion information for identifying second motion information used to generate a prediction signal of the second prediction region is restored from motion information candidates, and the second motion is based on the second motion information. The prediction information decoding means for restoring information And a storage means for storing motion information included in the restored prediction information, and the first prediction included in the target region based on the restored first motion information and the second motion information. A prediction signal generating means for generating a prediction signal for each of the area and the second prediction area, a residual signal restoring means for restoring the reproduction residual signal of the target area from the encoded data of the residual signal, and the prediction Recording means for restoring the pixel signal of the target area based on the signal and the reproduction residual signal, and storing the pixel signal as an already reproduced signal.
An image predictive decoding method according to the present invention is an image predictive decoding method executed by an image predictive decoding device, and is a target to be decoded from compressed data of an image divided into a plurality of regions and encoded. A data analysis step of extracting encoded data of prediction information instructing a prediction method used for prediction of a signal in a region and encoded data of a residual signal, and prediction for restoring motion information from the encoded data of the prediction information An information decoding step, wherein a prediction block partition type indicating the number of prediction regions into which the target region is subdivided is restored from the encoded data of the prediction information, and the prediction block partition type is the first target region The prediction region includes a second prediction region, the encoded data of the prediction information is further decoded, and the first prediction region is used to generate a prediction signal of the first prediction region. First merge identification information indicating whether or not to use a motion information candidate including at least one of decoded motion information and motion information indicating that the motion vector is 0 associated with an adjacent region adjacent to When the first merge identification information indicates that the motion information candidate is not used, the encoded data of the prediction information is further decoded to be used for generating a prediction signal of the first prediction region When the motion information is restored and the first merge identification information indicates that the motion information candidate is to be used, the encoded data of the prediction information is further decoded to generate the first prediction from the motion information candidate. The first selection information for specifying the first motion information used for generating the prediction signal of the region is restored, the first motion information is restored based on the first selection information, and the prediction information is encoded. Further recovery of data Then, the second merge identification information indicating whether to use the decoded motion information attached to the adjacent region adjacent to the second prediction region for generating the prediction signal of the second prediction region is restored. In the case where the second merge identification information indicates that the decoded motion information is not used, the encoded data of the prediction information is further decoded to generate a second prediction region prediction signal. When the motion information is restored and the second merge identification information indicates that the decoded motion information is to be used, the encoded data of the prediction information is further decoded, and a plurality of adjacent to the second prediction region 2nd motion information used for generation of a prediction signal of the second prediction region is specified from decoded motion information candidates excluding the first motion information among the decoded motion information accompanying the adjacent region Restore the second selection information, and select the second The second motion information is restored based on the selection information of the prediction information decoding step, the storage step of saving the motion information included in the restored prediction information, the restored first motion information, and the Based on second motion information, a prediction signal generation step for generating a prediction signal for each of the first prediction region and the second prediction region included in the target region, and encoded data of the residual signal A residual signal restoration step for restoring a reproduction residual signal of the target area, a pixel signal of the target area is restored based on the prediction signal and the reproduction residual signal, and the pixel signal is stored as a reproduced signal Recording step.
The image predictive decoding program according to the present invention is a prediction for instructing a prediction method used for predicting a signal in a target region to be decoded from compressed data of an image encoded by dividing a computer into a plurality of regions. A data analysis unit that extracts encoded data of information and encoded data of a residual signal; and a prediction information decoding unit that restores motion information from the encoded data of the prediction information, the encoding of the prediction information A prediction block partition type indicating the number of prediction regions for subdividing the target region is restored from data, and the prediction block partition type includes the target region including a first prediction region and a second prediction region. The decoded motion information associated with an adjacent region adjacent to the first prediction region is further used to generate the prediction signal of the first prediction region by further decoding the encoded data of the prediction information. And first merge identification information indicating whether to use a motion information candidate including at least one of motion information indicating that the motion vector is 0, and the first merge identification information is the motion information candidate. Is used, the encoded data of the prediction information is further decoded to restore the first motion information used to generate the prediction signal of the first prediction region, and the first merge identification When the information indicates that the motion information candidate is used, the first motion is further decoded from the motion information candidate to generate the prediction signal of the first prediction region by further decoding the encoded data of the prediction information. First selection information for identifying information is restored, the first motion information is restored based on the first selection information, the encoded data of the prediction information is further decoded, and the second prediction information is restored. For generation of prediction signal of region The second merge identification information indicating whether or not to use the decoded motion information associated with the adjacent region adjacent to the second prediction region is restored, and the second merge identification information does not use the decoded motion information. The encoded data of the prediction information is further decoded to restore the second motion information used to generate the prediction signal of the second prediction region, and the second merge identification information has been decoded. When indicating that the motion information is to be used, the encoded data of the prediction information is further decoded, and the first of the decoded motion information associated with a plurality of adjacent regions adjacent to the second prediction region Second selection information for specifying the second motion information used for generating the prediction signal of the second prediction region is restored from the decoded motion information candidates excluding the motion information, and the second selection information is used as the second selection information. Restore the second motion information based on Included in the target region based on the prediction information decoding means, storage means for saving the motion information included in the restored prediction information, and the restored first motion information and the second motion information. Prediction signal generating means for generating prediction signals for each of the first prediction region and the second prediction region, and a residual signal for restoring the reproduction residual signal of the target region from the encoded data of the residual signal It is made to function as a restoring means and a recording means for restoring the pixel signal of the target area based on the prediction signal and the reproduction residual signal and saving the pixel signal as a reproduced signal.
The following aspects of the present invention are also conceivable. The first aspect of the present invention relates to image predictive coding.

The first aspect of the present invention relates to image predictive coding, and can be described as an image predictive coding apparatus, method, and program as follows.
The image predictive coding apparatus according to the present invention includes a region dividing unit that divides an input image into a plurality of regions, and a target region that is an encoding target divided by the region dividing unit is a first prediction region and a second prediction region. Subdividing into prediction regions, determining the number of prediction regions suitable for the target region and the prediction block partition type indicating the region shape, and generating a signal having a high correlation with the first and second prediction regions The first and second motion information to be obtained from the first prediction information is respectively predicted, and the prediction block division type, the first and second motion information, and the prediction signal for the first prediction region are generated. First merge identification information indicating whether to use decoded motion information associated with an adjacent region adjacent to one prediction region, and the second prediction region for generating a prediction signal of the second prediction region Decoded motion information associated with an adjacent region adjacent to Prediction information estimation means for obtaining prediction information including second merge identification information indicating whether or not to use decoded motion information excluding the first motion information, and prediction associated with the target region Prediction information encoding means for encoding information, and prediction for generating prediction signals for each of the first prediction region and the second prediction region based on the first motion information and the second motion information Generated by a signal generation unit, a residual signal generation unit that generates a residual signal based on a prediction signal and a pixel signal of each of the first prediction region and the second prediction region, and the residual signal generation unit. Residual signal encoding means for encoding the residual signal, residual signal restoration means for generating a reproduction residual signal by decoding encoded data of the residual signal, the prediction signal and the reproduction residual Based on the difference signal, restoration of the target area Generates a prime signal, comprising a recording means for storing the reconstruction pixel signal as the already reproduced signal.
An image predictive encoding method according to the present invention is an image predictive encoding method executed by an image predictive encoding device, and is divided by an area dividing step for dividing an input image into a plurality of areas and the area dividing step. The target region to be encoded is subdivided into a first prediction region and a second prediction region, a prediction block division type indicating the number and region shape of prediction regions suitable for the target region is determined, First and second motion information for obtaining a signal having a high correlation with the first and second prediction regions from the already reproduced signal is predicted, respectively, and the predicted block division type, the first and second First merge identification information indicating whether or not to use motion information and decoded motion information associated with an adjacent region adjacent to the first prediction region for generating a prediction signal of the first prediction region; Prediction of the second prediction region Second merge identification indicating whether or not to use decoded motion information excluding the first motion information among decoded motion information associated with an adjacent region adjacent to the second prediction region for generation of a signal A prediction information estimation step for obtaining prediction information including information, a prediction information encoding step for encoding prediction information associated with the target region, the first motion information, and the second motion information , Based on a prediction signal generation step of generating a prediction signal of each of the first prediction region and the second prediction region, and a prediction signal and a pixel signal of each of the first prediction region and the second prediction region A residual signal generating step for generating a residual signal, a residual signal encoding step for encoding the residual signal generated by the residual signal generating step, and decoding the encoded data of the residual signal Due to playback residuals A residual signal restoring step for generating a signal, a recording step for generating a restored pixel signal of the target area based on the prediction signal and the reproduced residual signal, and storing the restored pixel signal as the reproduced signal; ,including.
An image predictive coding program according to the present invention includes a region dividing unit that divides an input image into a plurality of regions, and a target region that is an encoding target divided by the region dividing unit is a first prediction region. Subdividing into second prediction regions, determining the number of prediction regions suitable for the target region and the prediction block partition type indicating the region shape, and outputting a signal highly correlated with the first and second prediction regions First and second motion information to be obtained from the already reproduced signal is predicted, and the prediction block division type, the first and second motion information, and the prediction signal of the first prediction region are generated. To generate first merge identification information indicating whether or not to use decoded motion information associated with an adjacent region adjacent to the first prediction region, and to generate a prediction signal of the second prediction region. To the adjacent region adjacent to the predicted region Prediction information estimation means for obtaining prediction information including second merge identification information indicating whether or not to use decoded motion information excluding the first motion information among the corresponding decoded motion information; Based on prediction information encoding means for encoding prediction information associated with the target region, and the first motion information and the second motion information, each of the first prediction region and the second prediction region Prediction signal generation means for generating a prediction signal, residual signal generation means for generating a residual signal based on the prediction signal and the pixel signal of each of the first prediction area and the second prediction area, and the residual Residual signal encoding means for encoding the residual signal generated by the signal generation means; residual signal restoration means for generating a reproduction residual signal by decoding encoded data of the residual signal; and Based on the prediction signal and the reproduction residual signal Wherein generating a decompressed pixel signal of the target area, a recording means for storing the reconstruction pixel signal as the already reproduced signal, to function as Te.
The second aspect of the present invention relates to image predictive decoding, and can be described as an image predictive decoding apparatus, method, and program as follows.
The image predictive decoding device according to the present invention encodes prediction information that indicates a prediction method used to predict a signal of a target region to be decoded from compressed data of an image divided and encoded into a plurality of regions. Data analysis means for extracting encoded data and encoded data of the residual signal, and prediction information decoding means for restoring motion information from the encoded data of the prediction information, from the encoded data of the prediction information, When the prediction block division type indicating the number of prediction regions into which the target region is subdivided is restored, and the prediction block division type indicates that the target region includes a first prediction region and a second prediction region Whether to further decode the encoded data of the prediction information and use the decoded motion information associated with an adjacent region adjacent to the first prediction region to generate a prediction signal of the first prediction region Indicate 1 merge identification information is restored, and when the first merge identification information indicates that the decoded motion information is not used, the encoded data of the prediction information is further decoded, and When the first motion information used to generate the prediction signal is restored, and the first merge identification information indicates that the decoded motion information is used, the encoded data of the prediction information is further decoded, Reconstructing first selection information identifying first motion information used for generating a prediction signal of the first prediction region from decoded motion information associated with a plurality of adjacent regions adjacent to the first prediction region; The first motion information is restored based on the first selection information, the encoded data of the prediction information is further decoded, and the second prediction region is used to generate a prediction signal of the second prediction region. Decoded associated with the adjacent region adjacent to When the second merge identification information indicating whether to use motion information is restored, and the second merge identification information indicates that the decoded motion information is not used, the encoded data of the prediction information is further Decoding, reconstructing the second motion information used for generating the prediction signal of the second prediction region, and indicating that the second merge identification information uses the decoded motion information, From the decoded motion information candidates excluding the first motion information among the decoded motion information associated with a plurality of adjacent regions adjacent to the second prediction region by further decoding the encoded data, the first The prediction information decoding that restores the second selection information that identifies the second motion information used to generate the prediction signal of the second prediction region and restores the second motion information based on the second selection information Included in the restored forecast information Each of the first prediction region and the second prediction region included in the target region based on the restored first motion information and the second motion information. Prediction signal generation means for generating a prediction signal of the target area, residual signal restoration means for restoring the reproduction residual signal of the target region from the encoded data of the residual signal, and the prediction signal and the reproduction residual signal Recording means for restoring the pixel signal of the target area based on the image and storing the pixel signal as an already reproduced signal.
An image predictive decoding method according to the present invention is an image predictive decoding method executed by an image predictive decoding device, and is a target to be decoded from compressed data of an image divided into a plurality of regions and encoded. A data analysis step of extracting encoded data of prediction information instructing a prediction method used for prediction of a signal in a region and encoded data of a residual signal, and prediction for restoring motion information from the encoded data of the prediction information An information decoding step, wherein a prediction block partition type indicating the number of prediction regions into which the target region is subdivided is restored from the encoded data of the prediction information, and the prediction block partition type is the first target region The prediction area and the second prediction area, the encoded data of the prediction information is further decoded to generate a prediction signal of the first prediction area. When restoring the first merge identification information indicating whether to use the decoded motion information associated with the adjacent adjacent region, and indicating that the first merge identification information does not use the decoded motion information, The encoded data of the prediction information is further decoded to restore the first motion information used for generating the prediction signal of the first prediction region, and the first merge identification information uses the decoded motion information. The encoded data of the prediction information is further decoded, and from the decoded motion information associated with a plurality of adjacent regions adjacent to the first prediction region, the prediction signal of the first prediction region The first selection information for specifying the first motion information used for generation is restored, the first motion information is restored based on the first selection information, and the encoded data of the prediction information is further decoded. Of the second prediction region Second merge identification information indicating whether or not decoded motion information associated with an adjacent area adjacent to the second prediction area is used for measurement signal generation is restored, and the second merge identification information is decoded. In the case where it is indicated that the already-used motion information is not used, the encoded data of the prediction information is further decoded to restore the second motion information used for generating the prediction signal of the second prediction region, When the merge identification information indicates that the decoded motion information is used, the encoded motion data of the prediction information is further decoded, and the decoded motion information associated with a plurality of adjacent regions adjacent to the second prediction region Second selection information for identifying second motion information used for generating a prediction signal of the second prediction region is restored from decoded motion information candidates excluding the first motion information, Based on the second selection information, the second Based on the prediction information decoding step for restoring the motion information, the storage step for saving the motion information included in the restored prediction information, and the restored first motion information and the second motion information, A prediction signal generating step for generating a prediction signal of each of the first prediction region and the second prediction region included in the target region; and a reproduction residual signal of the target region is restored from the encoded data of the residual signal A residual signal restoring step, and a recording step of restoring the pixel signal of the target region based on the prediction signal and the reproduction residual signal and storing the pixel signal as an already reproduced signal.
The image predictive decoding program according to the present invention is a prediction for instructing a prediction method used for predicting a signal in a target region to be decoded from compressed data of an image encoded by dividing a computer into a plurality of regions. A data analysis unit that extracts encoded data of information and encoded data of a residual signal; and a prediction information decoding unit that restores motion information from the encoded data of the prediction information, the encoding of the prediction information A prediction block partition type indicating the number of prediction regions for subdividing the target region is restored from data, and the prediction block partition type includes the target region including a first prediction region and a second prediction region. The decoded motion information associated with an adjacent region adjacent to the first prediction region for further decoding the encoded data of the prediction information when generating the prediction signal of the first prediction region. When the first merge identification information indicating whether or not to use is restored, and the first merge identification information indicates that the decoded motion information is not used, the encoded data of the prediction information is further decoded. When the first motion information used for generating the prediction signal of the first prediction region is restored and the first merge identification information indicates that the decoded motion information is used, the prediction information is encoded. The data is further decoded to identify first motion information used for generating a prediction signal of the first prediction region from decoded motion information associated with a plurality of adjacent regions adjacent to the first prediction region. 1 selection information is restored, the first motion information is restored based on the first selection information, the encoded data of the prediction information is further decoded, and the prediction signal of the second prediction region Neighbors adjacent to the second prediction region for generation When the second merge identification information indicating whether to use the decoded motion information associated with the region is restored, and the second merge identification information indicates that the decoded motion information is not used, the prediction information When the encoded data is further decoded to restore the second motion information used to generate the prediction signal of the second prediction region, and the second merge identification information indicates that the decoded motion information is used In addition, decoded motion information excluding the first motion information among decoded motion information associated with a plurality of adjacent regions adjacent to the second prediction region by further decoding the encoded data of the prediction information Second selection information for identifying the second motion information used for generating the prediction signal of the second prediction region is restored from the candidates, and the second motion information is restored based on the second selection information. The prediction information decoding means to be restored; Based on the restored first motion information and the second motion information, storage means for storing motion information included in the original prediction information, and the first prediction region and the first prediction region included in the target region Prediction signal generation means for generating a prediction signal of each of the two prediction areas, residual signal restoration means for restoring the reproduction residual signal of the target area from the encoded data of the residual signal, the prediction signal, Based on the reproduction residual signal, the pixel signal of the target area is restored, and the pixel signal is made to function as a recording unit for storing the pixel signal as an already reproduced signal.
The following aspects of the present invention are also conceivable. The first aspect of the present invention relates to image predictive coding.

  An image predictive coding apparatus according to the first aspect of the present invention includes a region dividing unit that divides an input image into a plurality of regions, and a target region that is an encoding target divided by the region dividing unit. Motion information for obtaining a signal having a high correlation with the plurality of prediction regions from the already-reproduced signal by determining a prediction block division type indicating the number of prediction regions and the shape of the region suitable for the target region. Prediction information estimation means for obtaining prediction information including the prediction block division type and the motion information, prediction information encoding means for encoding prediction information associated with the target area, and associated with the target area Prediction signal generating means for generating a prediction signal for the target region based on prediction information to be generated, and a residual signal for generating a residual signal based on the prediction signal for the target region and the pixel signal for the target region. Generating means, residual signal encoding means for encoding the residual signal generated by the residual signal generating means, and residual for generating a reproduction residual signal by decoding the encoded data of the residual signal. Difference signal restoring means; and a recording means for generating a restored pixel signal of the target area by adding the prediction signal and the reproduced residual signal, and storing the restored pixel signal as the reproduced signal. To do. The prediction information encoding means has prediction information storage means for storing encoded prediction information, encodes the prediction block division type of the target region, and stores the prediction block division type in the prediction information storage means Then, based on the prediction information of the adjacent region adjacent to the target region, the number of encoded prediction regions in the target region, and the encoded prediction information of the target region, the target prediction region that is the next prediction region Motion information candidates used for generating a prediction signal are selected from the encoded motion information in the region adjacent to the target prediction region, and the selected motion information candidates are used according to the number of motion information candidates selected. Merge block information for instructing generation of a prediction signal of the target prediction area and motion information detected by the prediction information estimation means, or either the merge block information or the motion information Encoded and stores the motion information used for generating the predicted signal of the target prediction region in the prediction information storage unit.

  The image predictive coding method according to the first aspect of the present invention includes a region dividing step for dividing an input image into a plurality of regions, and a target region to be encoded divided by the region dividing step as a plurality of prediction regions. Motion information for obtaining a signal having a high correlation with the plurality of prediction regions from the already-reproduced signal by determining a prediction block division type indicating the number of prediction regions and the shape of the region suitable for the target region. Prediction information estimation step for obtaining prediction information including the prediction block division type and the motion information, a prediction information encoding step for encoding prediction information associated with the target region, and an associated with the target region A prediction signal generating step for generating a prediction signal of the target region based on the prediction information to be performed, and a residual based on the prediction signal of the target region and the pixel signal of the target region A residual signal generating step for generating a signal, a residual signal encoding step for encoding the residual signal generated by the residual signal generating step, and reproduction by decoding the encoded data of the residual signal A residual signal restoration step for generating a residual signal, and a restored pixel signal for the target region is generated by adding the prediction signal and the reproduced residual signal, and the restored pixel signal is stored as the reproduced signal. Recording step. In the prediction information encoding step, the prediction block division type of the target region is encoded, the prediction block division type is stored in a prediction information storage unit that stores the encoded prediction information, and adjacent to the target region. Based on the prediction information of the region, the number of encoded prediction regions in the target region, and the encoded prediction information of the target region, the motion information used for generating the prediction signal of the target prediction region that is the next prediction region A candidate is selected from the encoded motion information of the area adjacent to the target prediction area, and the prediction signal of the target prediction area using the selected motion information candidates is selected according to the number of selected motion information candidates. The merge block information instructing generation and the motion information detected in the prediction information estimation step, or the merge block information and the motion information are encoded, and the target The motion information used for generating the predicted signal measurement region is stored in the prediction information storage unit.

  Also, the image predictive coding program according to the first aspect of the present invention causes a computer to function as each unit of the image predictive coding device described above.

  According to the first aspect of the present invention, motion information candidates used for generating a prediction signal of a target prediction block are encoded prediction information (motion information and prediction block) of the target coding block and surrounding coding blocks. Therefore, the generation of inefficient codes is suppressed.

  In one embodiment, based on the number of encoded prediction regions in the target region, the prediction block partition type of the target region, and the prediction book partition type of an adjacent region adjacent to the target region, the next prediction region A candidate for motion information in a certain target prediction region can be selected from encoded motion information in a region adjacent to the target prediction region.

  In one embodiment, based on the number of encoded prediction regions in the target region and the prediction block division type of the target region, motion information candidates of the target prediction region that is the next prediction region are the target prediction. When selected from encoded motion information of a region adjacent to the region, the target region is divided into two prediction regions, and the target prediction region is a prediction region encoded second in the target region The motion information of a region adjacent to the target prediction region and not included in the target region can be selected as a candidate of motion information used for generating a prediction signal of the target prediction region.

  In one embodiment, the number of encoded prediction regions in the target region, the prediction block division type of the target region, the encoded motion information in the target region, and the adjacent regions adjacent to the target region. Based on the motion information, motion information candidates used for generating a prediction signal of the target prediction region which is the next prediction region are selected from the encoded motion information of the region adjacent to the target prediction region, and the target region Is divided into two prediction regions, the target prediction region is a prediction region encoded second in the target region, and motion information of the prediction region encoded first of the target region is When the motion information of the region adjacent to the target prediction region and not included in the target region is the same, the motion information of the region adjacent to the target prediction region is used to generate a prediction signal of the target prediction region It is determined that there is no motion information can be encoded.

  The second aspect of the present invention relates to image predictive decoding.

  The image predictive decoding device according to the second aspect of the present invention indicates a prediction method used for predicting a signal of a target region to be decoded from compressed data of an image divided and encoded into a plurality of regions. Encoded data of the prediction information, encoded data of the prediction signal of the target area, encoded data of the residual signal, decoding the encoded data of the prediction information, A prediction block division type indicating the number and a shape of a plurality of prediction regions, which are sub-regions of the target region, and a prediction information decoding unit that restores motion information for acquiring a prediction signal of each prediction region from the already reproduced signal; A prediction signal generating means for generating a prediction signal for the target area based on prediction information associated with the target area; and a residual for restoring the reproduction residual signal of the target area from the encoded data of the residual signal Signal recovery And means, to restore the pixel signal of the target region by adding the above-mentioned prediction signal and the reproduced residual signal, the pixel signal including a recording means for storing as the already reproduced signal. The prediction information decoding unit includes a prediction information storage unit that stores the decoded prediction information, decodes the prediction block division type of the target region, stores the prediction block division type in the prediction information storage unit, and is adjacent to the target region. Based on the prediction information of the region, the number of decoded prediction regions in the target region, and the decoded prediction information of the target region, motion information candidates used for generating a prediction signal of the target prediction region that is the next prediction region are obtained. Instructed to generate a prediction signal of the target prediction area using the selected motion information candidates according to the number of motion information candidates selected from the decoded motion information of the area adjacent to the target prediction area Decoding one of the merge block information and motion information to be performed, or the merge block information and the motion information, and storing the motion information used for generating a prediction signal of the target prediction region as the prediction information To save on stage.

  The image predictive decoding method according to the second aspect of the present invention indicates a prediction method used for predicting a signal of a target region to be decoded from compressed data of an image divided and encoded into a plurality of regions. A data analysis step for extracting encoded data of the prediction information, encoded data of the prediction signal of the target region, and encoded data of the residual signal; decoding the encoded data of the prediction information; A prediction block division type indicating the number and a shape of a plurality of prediction regions which are sub-division regions of the target region, and a prediction information decoding step for restoring motion information for obtaining a prediction signal of each prediction region from the already reproduced signal; A prediction signal generation step of generating a prediction signal of the target region based on prediction information associated with the target region, and a reproduction residual signal of the target region is restored from the encoded data of the residual signal Including a residual signal restoration step restores the pixel signal of the target region by adding the above-mentioned prediction signal and the reproduced residual signal, and a recording step of storing the pixel signal as the already reproduced signal. In the prediction information decoding step, the prediction block division type of the target region is decoded, the prediction block division type is stored in a prediction information storage unit that stores the decoded prediction information, and an adjacent region adjacent to the target region Motion information candidates used for generating a prediction signal of the target prediction area that is the next prediction area based on the prediction information of the target area, the number of decoded prediction areas in the target area, and the decoded prediction information of the target area. Instructing generation of a prediction signal of the target prediction region using the selected motion information candidates according to the number of motion information candidates selected from the decoded motion information of the region adjacent to the target prediction region One of the merge block information and the motion information, or the merge block information and the motion information is decoded, and the motion information used for generating the prediction signal of the target prediction region is the prediction To save the broadcast storage means.

  The image predictive decoding program according to the second aspect of the present invention causes a computer to function as each unit of the image predictive decoding device described above.

  According to the second aspect of the present invention, an image can be decoded from the compressed data generated by the above-described image predictive coding.

  In one embodiment, the next prediction region is based on the number of decoded prediction regions in the target region, the prediction block partition type of the target region, and the prediction book partition type of an adjacent region adjacent to the target region. Candidates for motion information in the target prediction region can be selected from the decoded motion information in the region adjacent to the target prediction region.

  In one embodiment, based on the number of decoded prediction regions in the target region and the prediction block division type of the target region, motion information candidates used for generating a prediction signal of the target prediction region that is the next prediction region Is selected from decoded motion information in a region adjacent to the target prediction region, the target region is divided into two prediction regions, and the target prediction region is decoded second in the target region. When it is a region, motion information of a region adjacent to the target prediction region and not included in the target region can be selected as a candidate for motion information of the target prediction region.

  In one embodiment, the number of decoded prediction areas in the target area, the prediction block division type of the target area, the decoded motion area in the target area, and motion information of adjacent areas adjacent to the target area Based on the above, motion information candidates used for generating a prediction signal of the target prediction region that is the next prediction region are selected from the decoded motion information of the region adjacent to the target prediction region, and the target region is divided into two The prediction region is divided into prediction regions, and the target prediction region is a prediction region that is decoded secondly in the target region, and the motion information of the prediction region that is decoded first in the target region is the target prediction region And the motion information of the region adjacent to the target prediction region is determined not to be used for generating the prediction signal of the target prediction region. Te, motion information may be decoded.

  The third aspect of the present invention relates to video coding.

  A moving image encoding apparatus according to the third aspect includes a dividing unit, a small section generating unit, a motion detecting unit, a prediction signal generating unit, a motion predicting unit, a difference motion vector generating unit, a residual signal generating unit, an adding unit, and a storage. Means and encoding means. The dividing unit divides the input image in the moving image into a plurality of sections. The small section generation unit divides the processing target section generated by the dividing unit into a plurality of small sections, and generates shape information specifying the shape of the small section. The motion detection means detects a motion vector of the section to be processed. The prediction signal generation unit generates a prediction signal of the section to be processed from the already reproduced image signal using the motion vector detected by the motion detection unit. The motion predicting means is based on the shape information generated by the small section generating means and the motion vector of the processed partial area that is the previous section or the small section in the processing order from the processing target section. A predicted motion vector is generated. The difference motion vector generation means generates a difference motion vector based on the difference between the motion vector used for generating the prediction signal of the processing target section and the prediction motion vector. The residual signal generation means generates a residual signal based on the difference between the prediction signal and the pixel signal of the section to be processed. The adding means adds the residual signal and the prediction signal to generate a reproduced image signal. The storage means stores the reproduced image signal as an already reproduced image signal. The encoding unit encodes the residual signal generated by the residual signal generation unit, the differential motion vector generated by the difference vector generation unit, and the shape information generated by the small block generation unit to compress the compressed data. Is generated. When the processing target sub-partition in the processing target section does not contact the previous section in the processing order with respect to the processing target sub-partition, the motion prediction unit determines whether the processing target sub-partition and the processing target sub-partion Based on the motion vector of the processed partial area belonging to the area including the processing target subsection among the one side area and the other side area with respect to the extension line of the boundary with the other subsection, A predicted motion vector of a processing target small section is generated.

  The moving image encoding method according to the third aspect includes (a) a dividing step of dividing an input image from a moving image into a plurality of sections, and (b) dividing a processing target section generated in the dividing step into a plurality of small sections. Detected in a sub-partition generation step for generating shape information that divides into sections and identifies the shape of the sub-partition, (c) a motion detection step for detecting a motion vector of the target section, and (d) a motion detection step. A prediction signal generation step of generating a prediction signal of the processing target section from the already reproduced image signal using the motion vector obtained, and (e) shape information generated in the small section generation step, and processing from the processing target section A motion prediction step for generating a predicted motion vector of a processing target partition based on a motion vector of a processed partial region that is a previous partition or a small partition in order; and (f) a processing pair A difference motion vector generation step for generating a difference motion vector based on the difference between the motion vector used to generate the prediction signal of the section and the prediction motion vector; and (g) a prediction signal and a pixel signal of the processing target section A residual signal generating step for generating a residual signal based on the difference; (h) an adding step for adding the residual signal and the prediction signal to generate a reproduced image signal; and (i) an already reproduced image signal being reproduced. A storage step for storing as an image signal; (j) a residual signal generated in a residual signal generation step; a differential motion vector generated in a differential motion vector generation step; and shape information generated in a subsection generation step And an encoding step of generating compressed data by encoding. In the motion prediction step, if the processing target sub-partition in the processing target sub-part does not touch the previous sub-part in the processing order from the processing target sub-partition, the processing target sub-partition and the processing target sub-partition The processing target based on the motion vector of the processed partial area belonging to the area including the processing target subsection among the one side area and the other side area with respect to the extension line of the boundary between the subsection The predicted motion vectors of the small sections are generated.

  The moving picture coding program according to the third aspect causes a computer to function as each unit of the moving picture coding apparatus described above.

  Of the two areas defined by the boundary extension lines described above, an area including a subsection that is not in contact with the previous section in the processing order may have a movement similar to the movement of the subsection. Is expensive. Therefore, according to the third aspect, the accuracy of the predicted motion vector is improved, the value of the difference motion vector is reduced, and the motion vector is encoded with a small code amount. Therefore, encoding efficiency is improved.

  The fourth aspect of the present invention relates to video decoding.

  The moving picture decoding apparatus according to the fourth aspect includes decoding means, motion prediction means, vector addition means, prediction signal generation means, addition means, and storage means. The decoding means decodes the compressed data and specifies the reproduction residual signal of the processing target section in the image, the difference motion vector of the processing target section, and the shapes of a plurality of small sections in the processing target section. Generate shape information. The motion prediction means generates a predicted motion vector of the processing target section based on the shape information and the motion vector of the processed partial area that is a section or a small section preceding the processing target section in the processing order. The vector addition unit adds the predicted motion vector generated by the motion prediction unit and the difference motion vector generated by the decoding unit to generate a motion vector of the processing target partition. The prediction signal generation means generates a prediction signal for the processing target section from the already reproduced image signal based on the motion vector of the processing target section. The adding means adds the prediction signal and the reproduction residual signal generated by the decoding means to generate a reproduced image signal. The storage means stores the reproduced image signal as an already reproduced image signal. When the processing target sub-partition in the processing target section does not contact the previous section in the processing order from the processing target sub-partition, the motion prediction means Based on the motion vector of the processed partial area belonging to the area including the processing target subsection among the one side area and the other side area with respect to the extension line of the boundary with the subsection, the processing target A predicted motion vector of a small section is generated.

  A moving image decoding method according to a fourth aspect is a method of generating a moving image by decoding compressed data, and (a) decoding residual data of a processing target section in the image by decoding the compressed data A decoding step for generating a differential motion vector of the processing target section, shape information for specifying the shapes of a plurality of small sections in the processing target section, (b) shape information, and processing order from the processing target section A motion prediction step for generating a predicted motion vector of a processing target partition based on a motion vector of a processed partial area that is a previous partition or a small partition, and (c) a predicted motion vector generated in the motion prediction step And a vector addition step for generating the motion vector of the processing target section by adding the difference motion vectors generated in the decoding step, and (d) based on the motion vector of the processing target section, A prediction signal generation step for generating a prediction signal of the target section from the already reproduced image signal, and (e) adding the prediction signal and the reproduction residual signal generated in the decoding step to generate a reproduction image signal And (f) a storage step of storing the reproduced image signal as an already reproduced image signal. In the motion prediction step, if the processing target sub-partition in the processing target sub-part does not touch the previous sub-part in the processing order from the processing target sub-partition, the processing target sub-partition and the processing target sub-partition The processing target based on the motion vector of the processed partial area belonging to the area including the processing target subsection among the one side area and the other side area with respect to the extension line of the boundary between the subsection The predicted motion vectors of the small sections are generated.

  The moving picture decoding program according to the fourth aspect causes a computer to function as each unit of the moving picture decoding apparatus described above.

  According to the fourth aspect, the predicted motion vector of the small section is generated from the decoded motion vector of the area including the small section that is not in contact with the previous section in the processing order. The predicted motion vector is likely to be similar to the motion vector of the small section. Therefore, according to the above-described embodiment, the accuracy of the predicted motion vector is improved, the value of the difference motion vector is reduced, and decoding from compressed data with a small bit amount is possible. Therefore, efficient decoding is realized.

  According to an image predictive encoding device, an image predictive encoding method, an image predictive encoding program, an image predictive decoding device, an image predictive decoding method, and an image predictive decoding program according to some aspects of the present invention, prediction of a target prediction block Since motion information candidates used for signal generation can be selected based on surrounding encoded or decoded information, there is an effect that motion information can be encoded more efficiently.

  Further, according to some other aspects of the present invention, a moving picture coding apparatus, a moving picture coding method, and a moving picture coding program capable of improving coding efficiency are provided. In addition, a moving picture decoding apparatus, a moving picture decoding method, and a moving picture decoding program corresponding to these moving picture encodings are provided.

It is a block diagram which shows the image predictive coding apparatus which concerns on one Embodiment. It is a schematic diagram for demonstrating the candidate of the motion information in the conventional block merging. It is a schematic diagram for demonstrating the candidate of the motion information in the block merging which concerns on one Embodiment. It is a flowchart explaining the process of the prediction information encoder shown in FIG. It is a flowchart which shows the procedure of the image prediction encoding method of the image prediction encoding apparatus shown in FIG. It is a block diagram which shows the image prediction decoding apparatus which concerns on one Embodiment. It is a flowchart explaining the process of the prediction information decoder shown in FIG. It is a flowchart which shows the procedure of the image prediction decoding method of the image prediction decoding apparatus shown in FIG. It is a 1st schematic diagram for demonstrating the process which uses the motion information of the several adjacent block adjacent to a target prediction block as the motion information of a target prediction block. It is a 2nd schematic diagram for demonstrating the process which uses the motion information of the several adjacent block adjacent to a target prediction block as the motion information of a target prediction block. It is a flowchart explaining the process which uses the motion information of the some adjacent block adjacent to a target prediction block as the motion information of a target prediction block. It is a 3rd schematic diagram for demonstrating the process which uses the motion information of the several adjacent block adjacent to a target prediction block as the motion information of a target prediction block. It is a 2nd example of the flowchart explaining the process which utilizes the motion information of the several adjacent block adjacent to a target prediction block as motion information of a target prediction block. It is a 2nd example of the schematic diagram for demonstrating the candidate of the motion information in the block merging which concerns on one Embodiment. It is a 3rd example of the schematic diagram for demonstrating the candidate of the motion information in the block merging which concerns on one Embodiment. It is a block diagram which shows the program which can perform the image predictive coding method which concerns on one Embodiment. It is a block diagram which shows the program which can perform the image predictive decoding method which concerns on one Embodiment. It is a figure which shows the hardware constitutions of the computer for performing the program recorded on the recording medium. It is a perspective view of a computer for executing a program stored in a recording medium. It is a schematic diagram for demonstrating the prediction block division | segmentation type of an encoding block. It is a schematic diagram regarding the motion estimation process (A) and template matching process (B) in inter-screen prediction. It is a schematic diagram for demonstrating the conventional intra prediction method. It is a figure for demonstrating motion compensation prediction. It is a figure which shows schematically the structure of the moving image encoder which concerns on one Embodiment. It is a figure for demonstrating the production | generation of a small division. It is a figure which shows the structure of the motion predictor in one Embodiment. It is a flowchart of the moving image encoding method which concerns on one Embodiment. It is a flowchart which shows the process of the motion predictor which concerns on one Embodiment. It is a figure which shows an example of the small block of an object block, and the surrounding partial area. It is a figure which shows another example of the small division of an object block, and the surrounding partial area. It is a figure which shows another example of the small division of an object block, and the surrounding partial area. It is a figure which shows another example of the small division of an object block, and the surrounding partial area. It is a figure which shows another example of the small division of an object block, and the surrounding partial area. It is a figure which shows schematically the structure of the moving image decoding apparatus which concerns on one Embodiment. It is a figure which shows the structure of the motion predictor which concerns on one Embodiment. It is a flowchart of the moving image decoding method which concerns on one Embodiment. It is a flowchart which shows the process of the motion predictor which concerns on one Embodiment. It is a figure which shows the structure of the moving image encoding program which concerns on one Embodiment. It is a figure which shows the structure of the moving image decoding program which concerns on one Embodiment.

  Hereinafter, various embodiments will be described in detail with reference to the accompanying drawings. In the description of the drawings, the same or equivalent elements are denoted by the same reference numerals, and redundant description is omitted.

  FIG. 1 is a block diagram illustrating an image predictive encoding device 100 according to an embodiment. The image predictive coding apparatus 100 includes an input terminal 101, a block divider 102, a prediction signal generator 103, a frame memory 104, a subtractor 105, a transformer 106, a quantizer 107, an inverse quantizer 108, and an inverse transformer. 109, an adder 110, a quantized transform coefficient encoder 111, an output terminal 112, a prediction block division type selector 113, a motion information estimator 114, a prediction information memory 115, and a prediction information encoder 116.

  The converter 106, the quantizer 107, and the quantized transform coefficient encoder 111 function as residual signal encoding means, and the inverse quantizer and inverse transformer function as residual signal restoring means. The prediction block division type selector 113 and the motion information estimator 114 function as prediction information estimation means, and the prediction information memory 115 and the prediction information encoder 116 function as prediction information encoding means.

  The input terminal 101 is a terminal for inputting a moving image signal composed of a plurality of images.

  The block divider 102 divides an image to be encoded, which is represented by a signal input from the input terminal 101, into a plurality of regions (encoded blocks). In the present embodiment, the image to be encoded is divided into blocks each having 16 × 16 pixels, but may be divided into blocks having other sizes or shapes. Further, blocks of different sizes may be mixed in the screen.

  The prediction block division type selector 113 divides the target region (target encoded block) to be subjected to the encoding process into prediction regions in which the prediction process is performed. For example, one of (A) to (H) in FIG. 20 is selected for each coding block, and the coding block is subdivided. The divided area is referred to as a prediction area (prediction block), and each of the division methods (A) to (H) in FIG. 20 is referred to as a prediction block division type. The method for selecting the prediction block division type is not limited. For example, each subdivision is performed on the signal of the target coding block input via the line L102, and prediction processing and coding processing described later are actually performed to reproduce the original signal of the coding block. For example, a method of selecting a division type that minimizes the rate distortion value calculated from the coding error signal power between the signals and the code amount required for coding the coding block can be used. The prediction block division type of the target coding block is output to the prediction information memory 115, the motion information estimator 114, and the prediction signal generator 103 via the line L113a, the line L113b, and the line L113c, respectively.

  The motion information estimator 114 detects motion information necessary for generating a prediction signal of each prediction block in the target coding block. The method of generating a prediction signal (prediction method) is not limited, but inter-screen prediction and intra-screen prediction (not shown for intra-screen prediction) as described in the background art are applicable. Here, it is assumed that motion information is detected by block matching shown in FIG. The original signal of the target prediction block to be predicted can be generated from the original signal of the encoded block input via the line L102a and the prediction block division type of the target encoded block input via the line L113b. A prediction signal having a minimum absolute value error sum with respect to the original signal of the target prediction block is detected from the image signal acquired via the line L104. In this case, the motion information includes a motion vector, an inter-screen prediction mode (forward / backward / bidirectional), a reference screen number, and the like. The detected motion information is output to the prediction information memory 115 and the prediction information encoder 116 via the line L114.

  The prediction information memory 115 stores the input motion information and the prediction block division type.

  The prediction information encoder 116 selects motion information candidates used for block merging of each prediction block, entropy-encodes the prediction information of the target coding block, and outputs the encoded data to the output terminal 112 via the line L116. To do. The entropy encoding method is not limited, but arithmetic encoding, variable length encoding, and the like can be applied. The prediction information includes block merging information for performing block merging using the motion information of the block adjacent to the prediction block, in addition to the prediction block division type of the target coding block and the motion information of the prediction block. It is. The processing of the prediction information encoder 116 will be described later.

  In the prediction signal generator 103, based on the motion information of each prediction block in the target coding block input via the line L114 and the prediction block division type input via the line L113c, the reproduced signal is received from the frame memory 104. Obtaining and generating a prediction signal of each prediction block in the target coding block.

  The prediction signal generated by the prediction signal generator 103 is output to the subtracter 105 and the adder 110 via the line L103.

  The subtractor 105 subtracts the prediction signal for the target coding block input via the line L103 from the pixel signal of the target coding block divided by the block divider 102 and input via the line L102b. Generate a residual signal. The subtractor 105 outputs the residual signal obtained by the subtraction to the converter 106 via the line L105.

  The converter 106 is a part that performs discrete cosine transform on the input residual signal. The quantizer 107 is a part that quantizes the transform coefficient that has been discrete cosine transformed by the transformer 106. The quantized transform coefficient encoder 111 entropy codes the transform coefficient quantized by the quantizer 107. The encoded data is output to the output terminal 112 via the line L111. The entropy encoding method is not limited, but arithmetic encoding, variable length encoding, and the like can be applied.

  The output terminal 112 collectively outputs information input from the prediction information encoder 116 and the quantized transform coefficient encoder 111 to the outside.

  The inverse quantizer 108 inversely quantizes the quantized transform coefficient. The inverse transformer 109 restores the residual signal by inverse discrete cosine transform. The adder 110 adds the restored residual signal and the prediction signal input via the line L103, reproduces the signal of the target coding block, and stores the reproduced signal in the frame memory 104. In the present embodiment, the converter 106 and the inverse converter 109 are used, but other conversion processes in place of these converters may be used. Further, the converter 106 and the inverse converter 109 are not essential. Thus, in order to use for the prediction signal production | generation of a subsequent object encoding block, the reproduced signal of the object encoding block encoded is decompress | restored by reverse processing, and is memorize | stored in the frame memory 104. FIG.

  Next, the process of the prediction information encoder 116 will be described. First, the prediction information encoder 116 selects motion information candidates used for block merging of each prediction block (motion information candidates used for prediction signal generation in the target prediction region) from the motion information of blocks adjacent to the target prediction block. To do. Block merging refers to generating a prediction signal of a target prediction block using motion information of adjacent blocks. Next, the prediction information encoder 116 compares the motion information detected by the motion information estimator 114 with the selected candidate motion information, and determines whether or not block merging can be performed. Then, the prediction information encoder 116 entropy codes the block merging information and / or motion information together with the prediction block division type according to the number of motion information candidates used for block merging and whether or not the block merging can be performed. Turn into. The block merging information is adjacent to the target prediction block and merge identification information (merge_flag) indicating whether or not to generate a prediction signal of the target prediction block using motion information of adjacent blocks, that is, whether or not to perform block merging. It is composed of merge block selection information (merge_flag_left) indicating which one of the motion information of the two blocks is used to generate the prediction signal of the target prediction block.

  When there are no motion information candidates used for block merging of each prediction block, these two pieces of information, that is, merge identification information and merge block selection information do not need to be encoded. When there is one motion information candidate, merge identification information is encoded. When there are two or more motion information candidates and block merging is performed, two pieces of information, namely merge identification information and merge Encode block selection information. Even if the number of motion information candidates is two or more, if block merging is not performed, the merge block selection information need not be encoded.

  FIG. 3 is a schematic diagram for explaining a motion information candidate selection process used for block merging of a prediction block according to an embodiment. FIG. 3 shows an example of a prediction block division type in which an encoded block is divided into two vertically (divided into two blocks on the left and right), similarly to the block 301 shown in FIG. Hereinafter, the block 301 will be described as an example, but the same description can be applied to the blocks 302, 304, 305, 306, and 307.

Selection of motion information candidates is performed based on the following information.
1) Number of encoded / decoded prediction blocks in the target coding block 2) Prediction block division type of the target coding block 3) Prediction block division type of a block adjacent to the target prediction block 4) In the target coding block 5) Motion information and prediction mode of the block adjacent to the target prediction block (intra prediction / inter prediction)

  In the example of FIG. 3, candidates for motion information used for block merging are selected using the information of 1), 2), 4), and 5).

  First, it can be seen from the information 2) that the total number of prediction blocks in the target coding block 400 is two prediction blocks T1 and T2, and the coding block is vertically divided into two. Whether the next prediction block is the prediction block T1 or the prediction block T2 can be known from the information 1).

  When the next prediction block is the prediction block T1 (the number of encoded / decoded prediction blocks in the target encoding block is 0), the motion information of the adjacent block A and the adjacent block B is a candidate for block merging motion information. (The arrows in the drawing indicate that the motion information of adjacent blocks A and B is a candidate of motion information used for generating a prediction signal of the prediction block T1). At this time, if the adjacent block A or B is a block generated by intra prediction or a block outside the screen, the motion information of the block may be excluded from the motion information candidates for block merging. (Pseudo motion information can be set as a default value. For example, a motion vector is set to 0 and a reference screen number is set to 0). If the motion information of two adjacent blocks A and B match, the motion information of one adjacent block may be excluded from the candidates.

  When the next prediction block is the prediction block T2 (the number of encoded / decoded prediction blocks in the target coding block is 1), as shown in FIG. The motion information is excluded from candidates for block merging motion information. This is because the prediction block T1 and the prediction block T2 are divided into two assuming that a prediction signal is originally generated with different motion information. That is, this is to avoid that the motion information of the prediction block T1 and the motion information of the prediction block T2 are the same motion information. By this processing, the block merging motion information of the prediction block T2 becomes one, so that the cost required for encoding the merge block selection information can be reduced (the arrow in the drawing indicates that the motion information of the adjacent block D is the prediction block T2). Can be used to generate prediction signals for

  Further, based on the information of 4) and 5), the motion information of the prediction block T1 and the adjacent block D is compared, and when the motion information of the prediction block T1 and the adjacent block D matches, ), The motion information of the adjacent block D is also excluded from the block merging motion information candidates. This is because when the prediction signal of the prediction block T2 is generated using the motion information of the adjacent block D, the motion information of the prediction blocks T1 and T2 is the same. With this process, the motion information of block merging of the prediction block T2 becomes zero, and the cost required for encoding the merge identification information and the merge block selection information can be reduced.

  FIG. 4 is a flowchart of the prediction information encoder 116 that implements the processing of FIG.

  First, the prediction information encoder 116 encodes the prediction block division type of the target coding block, and stores the prediction block division type in the prediction information memory 115. At the same time, the prediction information encoder 116 sets the prediction block number N in the target encoded block based on the encoded prediction block division type, and resets the target prediction block number i to 0 (step S151). Next, the prediction information encoder 116 determines whether the target prediction block is the prediction block that is encoded last in the target coding block, and the number of prediction blocks in the target coding block is two or more. (Step S152). For example, in the case of N = 2, when i = 1, determination is possible, and the process proceeds to step S157. In the case of N = 4 ((D) in FIG. 20), the determination is possible when i = 3. If the determination is no, the process proceeds to step S153. In the case of FIG. 3, when the target prediction block is the prediction block T1, the process proceeds to step S153, and when the target prediction block is the prediction block T2, the process proceeds to step S157.

  In step S153, merge identification information is encoded. The merge identification information is acceptable when the motion information of the target prediction block matches the motion information candidate of block merging (merge_flag = 1, the prediction signal of the target prediction block is generated using the motion information candidate) If they do not match, the result is no (merge_flag = 0, the prediction signal of the target prediction block is generated using the encoded motion information). Next, when the motion information of the target prediction block matches the motion information candidate of block merging, the process proceeds to step S164. In step S164, the prediction information encoder 116 determines whether there are two motion information candidates. If the number of motion information candidates is two, the prediction information encoder 116 encodes merge block selection information and performs processing. Advances to step S155. If the number of motion information candidates is one, the process proceeds to step S165. On the other hand, when the motion information of the target prediction block does not match the motion information candidate for block merging, the process proceeds to step S156, and the prediction information encoder 116 encodes the motion information detected by the motion information estimator 114. Go to step S165.

  In step S157, the prediction information encoder 116 determines whether all the encoded motion information of the target encoded block matches the motion information of the adjacent block that does not belong to the target encoded block. The description of step S157 means that when N = 2, the motion information of the prediction block T1 and the adjacent block D shown in FIG. Further, in the description of step S157, when N = 4 ((D) in FIG. 20), the target prediction block is the lower right divided block, and the other three prediction blocks (upper left, upper right, and lower left). This means comparing the movement information of When the determination is possible (compared motion information matches), the number of motion information candidates used for block merging of the target prediction block is 0 as shown in the example of FIG. 3B. The prediction information encoder 116 encodes the motion information detected by the motion information estimator 114 without sending the block merging information, and the process proceeds to step S165 (step S160). On the other hand, if the determination is negative (compared motion information does not match), the process proceeds to step S163. When N = 4, the motion information of the upper right and lower left blocks in the target coding block becomes the adjacent block of the target prediction block. Therefore, applying block merging to the target prediction block (lower right) when the motion information of the three prediction blocks (upper left, upper right, lower left) matches is the prediction signal of the four prediction blocks in the target coding block. Means that all are generated with the same motion information. Therefore, when N = 4 and the motion information of the three prediction blocks (upper left, upper right, lower left) is the same, the number of motion information candidates of the target prediction block (lower right) is set to zero.

  In step S163, the prediction information encoder 116 determines whether or not the prediction block division type of the target encoded block is a two-division type. If the determination is negative, the process proceeds to step S153 (hereinafter described). Omitted). If the determination in step S163 is possible, the process proceeds to step S158, and the prediction information encoder 116 encodes the merge identification information. In this case, since the number of motion information candidates used for block merging of the target prediction block is one as in the example of FIG. 3A, encoding of merge block selection information is unnecessary. Next, when the motion information of the target prediction block matches the motion information candidate of block merging, the process proceeds to step S165. If the motion information of the target prediction block does not match the motion information candidate for block merging, the process proceeds to step S160, and the prediction information encoder 116 encodes the motion information detected by the motion information estimator 114. The process proceeds to step S165.

  In step S165, the motion information of the target block is stored in the prediction information memory 115. Subsequently, in step S161, the prediction information encoder 116 determines whether or not encoding of all prediction blocks in the target coding block has been completed (i = N−1), and i = N−. When 1, the prediction information encoding process of the target encoding block is finished. When i <N−1, i is updated in step S162 (i = i + 1), and the motion of the next prediction block The process returns to step S152 for the information encoding process.

In this manner, motion information candidates used for block merging of a prediction block can be selected in advance using the following information, so that block merging information can be efficiently transmitted.
1) Number of encoded / decoded prediction blocks in target coding block 2) Prediction block division type of target coding block 4) Motion information of encoded / decoded prediction block in target coding block 5) Motion information and prediction mode (intra-screen prediction / inter-screen prediction) of blocks adjacent to the target prediction block

  FIG. 5 is a flowchart showing a procedure of an image predictive encoding method in the image predictive encoding device 100 according to the present embodiment. First, the input image is divided into 16 × 16 encoded blocks by the block divider 102 (may be divided into blocks of other sizes or shapes. In addition, blocks of different sizes may be mixed in the screen. Good). Then, the prediction block division type selector 113 and the motion information estimator 114 determine the prediction block division type of the target coding block to be encoded and the motion information of each prediction block (step S101). Next, the prediction information encoder 116 encodes the prediction information (step S102, FIG. 4).

  Next, the prediction signal generator 103 generates a prediction signal of the target coding block based on the prediction block division type of the target coding block and the motion information of each prediction block, and the pixel signal of the target coding block and the prediction The residual signal indicating the difference from the signal is transform-coded by the transformer 106, the quantizer 107, and the quantized transform coefficient encoder 111 (step S103). Then, the prediction information and the encoded data of the quantized transform coefficient are output via the output terminal 112 (step S104).

  In order to predictively encode the subsequent target encoded block, the residual signal encoded after these processes or in parallel with these processes is decoded by the inverse quantizer 108 and the inverse transformer 109. Then, the adder 110 adds the decoded residual signal and the prediction signal to reproduce the signal of the target coding block. The reproduction signal is stored as a reference screen in the frame memory 104 (step S105). If all the target coding blocks have not been processed, the process returns to step S101, and processing for the next target coding block is performed. If all the target coding blocks have been processed, the process ends (step S106).

  Next, image predictive decoding according to an embodiment will be described. FIG. 6 is a block diagram illustrating an image predictive decoding device 200 according to an embodiment. The image predictive decoding apparatus 200 includes an input terminal 201, a data analyzer 202, an inverse quantizer 203, an inverse transformer 204, an adder 205, an output terminal 206, a quantized transform coefficient decoder 207, a prediction information decoder 208, A frame memory 104, a prediction signal generator 103, and a prediction information memory 115 are provided.

  The inverse quantizer 203, the inverse transformer 204, and the quantized transform coefficient decoder 207 function as residual signal decoding means. The decoding means by the inverse quantizer 203 and the inverse transformer 204 may be performed using other than these. Further, the inverse converter 204 may not be provided. The prediction information memory 115 and the prediction information decoder 208 function as prediction information decoding means.

  The input terminal 201 inputs compressed data that has been compression-encoded by the above-described image predictive encoding method. The compressed data includes information on quantized transform coefficients obtained by transform-quantizing an error signal and entropy-coding the encoded block divided into a plurality of pieces, and encoded data of prediction information for generating a prediction signal of the block. It is included. Here, the prediction information includes block merging information for performing block merging using motion information that is a candidate for block merging, in addition to the prediction block division type of the target coding block and the motion information of the prediction block. . The motion information includes a motion vector, an inter-screen prediction mode (forward / backward / bidirectional), a reference screen number, and the like.

  The data analyzer 202 analyzes the compressed data input to the input terminal 201, separates the target coding block to be decoded into encoded data of quantized transform coefficients and encoded data of prediction information, and outputs a line L202a. , Output to the quantized transform coefficient decoder 207 and the prediction information decoder 208, respectively, via the line L202b.

  The prediction information decoder 208 selects motion information candidates to be used for block merging of each prediction block, and entropy-decodes encoded data of prediction information associated with the target encoded block. The decoded prediction information is output to the prediction information memory 115 and the prediction signal generator 103 via the line L208a and the line L208b, respectively. The prediction information memory 115 stores the input prediction information. The processing of the prediction information decoder 208 will be described later.

  The prediction signal generator 103 acquires the already reproduced signal from the frame memory 104 based on the prediction information of the target coding block input via the line L208a, and generates a prediction signal of each prediction block in the target coding block. To do. The generated prediction signal is output to the adder 205 via the line L103.

  The quantized transform coefficient decoder 207 entropy-decodes the encoded data of the quantized transform coefficient of the residual signal in the target coding block, and outputs it to the inverse quantizer 203 via the line L207.

  The inverse quantizer 203 inversely quantizes the residual signal information of the target coding block input via the line L207. The inverse transformer 204 performs inverse discrete cosine transform on the inversely quantized data.

  The adder 205 adds the prediction signal generated by the prediction signal generator 103 to the residual signal restored by the inverse quantizer 203 and the inverse transformer 204, and outputs the reproduced pixel signal of the target coding block. The data is output to the output terminal 206 and the frame memory 104 via L205. The output terminal 206 outputs to the outside (for example, a display).

  The frame memory 104 stores the reproduction image output from the adder 205 as a reference screen as a reference reproduction image for the next decoding process.

  FIG. 7 is a flowchart of the prediction information decoder 208 that implements the processing of FIG.

  First, the prediction information decoder 208 decodes the prediction block division type of the target coding block and stores it in the prediction information memory 115. At the same time, the prediction information decoder 208 sets the prediction block number N in the target coding block based on the decoded prediction block division type, and resets the target prediction block number i to 0 (step S251). Next, the prediction information decoder 208 determines whether the target prediction block is the prediction block decoded last in the target coding block, and the number of prediction blocks in the target coding block is 2 or more (step) S252). For example, in the case of N = 2, when i = 1, determination is possible, and the process proceeds to step S258. In the case of N = 4 ((D) in FIG. 20), the determination is possible when i = 3. If the determination is no, the process proceeds to step S253. In FIG. 3, when the target prediction block is the prediction block T1, the process proceeds to step S253, and when the target prediction block is the prediction block T2, the process proceeds to step S258.

  In step S253, the merge identification information is decoded. Here, when the merge identification information is acceptable (merge_flag = 1), the merge identification information indicates that the prediction signal of the target prediction block is generated using the motion information candidates. On the other hand, when the merge identification information is NO (merge_flag = 0), it indicates that the prediction signal of the target prediction block is generated using the decoded motion information. In the next step S254, the prediction information decoder 208 determines whether the merge identification information indicates decoding of motion information, that is, whether the value of merge_flag is 0. If the decoded merge_flag value is 0, the prediction information decoder 208 decodes motion information for generating a prediction signal of the target prediction block (step S257), and the process proceeds to step S267. When the value of merge_flag is 1, the prediction information decoder 208 determines in step S266 whether the number of motion information candidates used for block merging is two, and when the number of candidates is two. Decodes the merge block selection information, and the process proceeds to step S256 (step S255). When the number of motion information candidates used for block merging of the target prediction block is one, the process proceeds to step S256. In step S256, when the number of motion information candidates is one, the prediction information decoder 208 determines the motion information as the motion information of the target prediction block. When the number of motion information candidates is two, the prediction information decoder 208 determines the motion information of the adjacent block indicated by the merge block selection information as the motion information of the target prediction block.

  In step S258, the prediction information decoder 208 determines whether all the decoded motion information of the target coding block matches the motion information of the neighboring block that does not belong to the target coding block. The description of step S258 means that when N = 2, the motion information of the prediction block T1 and the adjacent block D shown in FIG. In the description of step S258, when N = 4 ((D) in FIG. 20), the target prediction block is a lower right divided block, and the other three prediction blocks (upper left, upper right, lower left) ) Motion information. When the determination is possible (compared motion information matches), the number of motion information candidates used for block merging of the target prediction block is 0 as shown in the example of FIG. 3B. The prediction information decoder 208 decodes the motion information used for generating the prediction signal of the target prediction block without decoding the block merging information, and the process proceeds to step S267 (step S262). On the other hand, if the determination is negative (compared motion information does not match), the process proceeds to step S265. When N = 4, the motion information of the upper right and lower left blocks in the target coding block becomes the adjacent block of the target prediction block. Therefore, applying block merging to the target prediction block (lower right) when the motion information of the three prediction blocks (upper left, upper right, lower left) matches is the prediction signal of the four prediction blocks in the target coding block. Means that all are generated with the same motion information. Therefore, when N = 4 and the motion information of the three prediction blocks (upper left, upper right, and lower left) is the same, the number of motion information candidates of the target prediction block (lower right) is set to zero.

  In step S265, the prediction information decoder 208 determines whether the prediction block division type of the target coding block is a two-division type. If the determination is negative, the process proceeds to step S253 (hereinafter, description is omitted). ). If the determination in step S265 is possible, the process proceeds to step S259, and the prediction information decoder 208 decodes the merge identification information. In this case, since the number of motion information candidates used for block merging of the target prediction block is one as in the example of FIG. 3A, decoding of merge block selection information is unnecessary.

  In the next step S260, the prediction information decoder 208 determines whether the merge identification information indicates decoding of motion information, that is, whether the value of merge_flag is 0. When the decoded merge_flag value is 0, the prediction information decoder 208 decodes motion information for generating a prediction signal of the target prediction block (step S262), and the process proceeds to step S267. If the value of merge_flag is 1, the process proceeds to step S261. In step S261, since there is one candidate of motion information, as shown in FIG. 3A, the prediction information decoder 208 determines the motion information of the adjacent block D as the motion information of the target prediction block, The process proceeds to step S267.

  In step S267, the motion information of the restored prediction block is stored in the prediction information memory 115. Subsequently, in step S263, the prediction information decoder 208 determines whether or not decoding of all prediction blocks in the target coding block has been completed (i = N−1), and i = N−1. In this case, the prediction information decoding process for the target coding block is terminated. When i <N−1, i is updated in step S264 (i = i + 1), and motion information decoding for the next prediction block is performed. Return to step S252 for processing.

  Next, the image predictive decoding method in the image predictive decoding apparatus 200 shown in FIG. 6 will be described with reference to FIG. First, compressed data is input via the input terminal 201 (step S201). Then, the data analyzer 202 analyzes the data of the compressed data, and extracts prediction information regarding the target region to be decoded and encoded data of the quantized transform coefficient. The prediction information is decoded by the prediction information decoder 208 (S203).

  Thereafter, based on the restored prediction information, the prediction signal generator 103 generates a prediction signal of the target coding block (S204).

  The quantized transform coefficient decoded by the quantized transform coefficient decoder 207 is inversely quantized by the inverse quantizer 203 and inversely transformed by the inverse transformer 204 to generate a reproduction residual signal (S205). ). Then, the generated prediction signal and the reproduction residual signal are added to generate a reproduction signal, and this reproduction signal is stored in the frame memory 104 to reproduce the next target coding block (step S206). ). If there is the next compressed data, the process of S204 to S206 is repeated (S207), and all data is processed to the end.

  Up to this point, the number of adjacent blocks adjacent to the prediction block has been described with an example of 2 or less. Next, when the number of adjacent blocks adjacent to the upper and left block boundaries of the prediction block is 3 or more Focus on it.

  In the example of FIG. 3, there are two adjacent blocks in contact with the prediction block. However, depending on the combination of the encoded block and the prediction block division type of the adjacent block, the prediction block may be in contact with two or more adjacent blocks. Occur. FIG. 9 shows an example in which three adjacent blocks are in contact with the prediction block. Here, the block 301 in FIG. 20 will be described as an example, but the same description can be applied to the blocks 302, 304, 305, 306, and 307.

  In (A) and (B) of FIG. 9, the target coding block 400 has two prediction blocks obtained by vertically dividing the block 400 into two blocks, whereas the block 401 in contact with the left side of the prediction block T1. Is divided horizontally into two parts (divided into two upper and lower blocks). Therefore, the prediction block T1 is in contact with three adjacent blocks A, B, and C. In this case, if the encoding side and the decoding side are determined in advance so that the two adjacent blocks A and B in contact with the upper left vertex of the target prediction block are representative, the number of adjacent blocks can be limited to two. The techniques described above can be used.

  On the other hand, as shown in FIG. 9B, a method of virtually dividing the prediction block T1 horizontally into two in accordance with the prediction block division type of the adjacent block 401 can also be used. In this case, the target prediction block T1 is divided into T1a and T1b, and the prediction signal of the block T1a and the prediction signal of T1b are generated using two pieces of motion information belonging to the adjacent block A and the block C, respectively.

  At this time, if the selection candidates of merge block selection information are two combinations of the motion information of the adjacent block B in FIG. 9A and the motion information of the adjacent blocks A and C in FIG. The merge block selection information can be efficiently encoded without changing the configuration of the block merging information.

  On the other hand, if one of (A) in FIG. 9 and (B) in FIG. 9 is identified by the merge block selection information and (B) in FIG. 9 is selected, the second merge is further performed for each virtual block. The identification information may be transmitted to identify whether the prediction signal of the virtual block is generated based on the motion information of the adjacent block or whether the motion information is encoded / decoded.

Note that, without dividing the prediction block T1, the selection candidates of the merge block selection information in the prediction block T1 are the three pieces of motion information of the adjacent blocks A, B, and C, and the motion information used for generating the prediction signal of T1 is 3 You may make it select from an item, However In this case, the following change is required.
1. Before Step S164 in FIG. 4 and Step S266 in FIG. 7, a flow of “obtaining the prediction block division type of the adjacent block and deriving the number of adjacent blocks of the prediction block” is added.
2. Step S164 in FIG. 4 and step S266 in FIG. 7 are changed to “two or more pieces of selection candidate motion information”.
3. Merge block selection information is extended to information for selecting one from three or more candidates.

  The block merging process shown in FIGS. 9A and 9B can be realized by extending step S256 in FIG. 7 to the process shown in FIG. First, in step S256a, the prediction block division type of the encoded block that contacts the target prediction block is acquired. Next, in step S256b, the number M of prediction blocks adjacent to the block boundary indicated by the decoded merge block selection information is derived from the acquired prediction block division type. For example, in FIG. 9B, M = 2. In step S256c, it is determined whether the value of M is greater than 1 (M> 1). When M> 1, the target prediction block is divided into M virtual blocks, and the motion information of M adjacent blocks is set to the divided M virtual blocks (additionally, merge identification is performed for each virtual block) Information may be sent to determine whether to decode the motion information). When M = 1, the motion information of the adjacent block that is a candidate for block merging is set in the motion information of the target prediction block.

As described above, according to FIGS. 7 and 11, selection of motion information candidates in the example of FIG. 9 is performed based on the following information.
1) Number of encoded / decoded prediction blocks in the target coding block 2) Prediction block division type of the target coding block 3) Prediction block division type of a block adjacent to the target prediction block In the case where there are three or more candidates, the information 3) that is not used in selecting the motion information candidates according to the example of FIG. 3 is used.

  FIG. 9C illustrates an example in which a block adjacent to the left side of the prediction block 400 is asymmetrically divided into two. Even in this case, a method of virtually dividing the prediction block T1 horizontally into two in accordance with the prediction block division type of the adjacent block 401 (division into blocks T1a and T1b) can be used. That is, the prediction signal of the target prediction block T1 can be generated using the combination of the motion information of the adjacent blocks A and C in FIG. 9C as the candidate of the block merging motion information of the prediction block T1.

  Further, as shown in (D) to (F) of FIG. 9, the adjacent block 401 is used even when the prediction block division type of the encoded block is one type as shown in the block 300 of FIG. 20. In addition to the prediction block division type, a method of generating a prediction signal by virtually dividing the prediction block T1 (block 400) horizontally (dividing into a plurality of blocks arranged in the vertical direction) can be used. Further, although not illustrated, when the adjacent block 402 is vertically divided (divided into a plurality of blocks arranged in the horizontal direction), the prediction block T1 (block 400) is virtually combined with the prediction block division type of the adjacent block 402. A method of generating a prediction signal by vertically dividing into two can be used.

  In addition, even when an intra-screen prediction block (intra) is included in a block adjacent to the prediction block, a method of generating a prediction signal by virtually dividing the prediction block by setting a rule in advance can be used. (A) to (F) of FIG. 10 illustrate an example in which a plurality of adjacent blocks A, C, E, and G in contact with the left side of the prediction block include an intra-screen prediction block (intra). Based on the prediction block division type of the adjacent block and the prediction mode (inter-screen / in-screen) included in the prediction information, the intra-screen prediction block of the adjacent block is virtually integrated into the inter-screen prediction block with motion information (drawing) Thick line). In this example, the intra-screen prediction block is virtually integrated into an inter-screen prediction block that is close to the upper left vertex of the adjacent block and close. As a result, the prediction block T1 is virtually divided according to the number of inter-screen prediction blocks in adjacent blocks, as shown in (A) to (F) of FIG. As described above, even when an adjacent block includes an intra-screen prediction block (intra), a prediction signal can be generated by block merging using the motion information of the inter-screen prediction block in the adjacent block.

The rule for integrating the intra-screen prediction block of the adjacent block into the inter-screen prediction block is not limited. A plurality of such rules may be prepared, and the rules may be selected and encoded for each frame or each slice.
In this case, selection of motion information candidates is performed based on the following information.
1) Number of encoded / decoded prediction blocks in the target coding block 2) Prediction block division type of the target coding block 3) Prediction block division type of a block adjacent to the target prediction block 5) Adjacent to the target prediction block Block prediction mode (intra-screen prediction / inter-screen prediction)

  FIG. 12 shows an example in which the encoding block 400 and the adjacent block 402 are similarly divided into two vertically, but the division shapes are different. Also in this example, the prediction block T1 (block including the blocks T1a and T1b) in FIG. 12A and the prediction block T2 (block including the blocks T2a and T2b) in FIG. Have. For T1 in FIG. 12A, by applying the processing flow in FIG. 11 to step S256 in FIG. 7, the prediction block T1 is virtually divided into two vertically by T1a and T1b, and blocks Ba and Bb are respectively obtained. It is possible to perform block merging in which motion information is set. In addition, for T2 in FIG. 12B, the processing flow in FIG. 13 described below is applied to step S261 in FIG. 7, so that the prediction block T2 is virtually divided into two vertically by T2a and T2b. Thus, block merging in which motion information of the blocks Ba and Bb is set can be performed. At this time, the second merge identification information may be transmitted for each virtual block to identify whether a prediction signal of the virtual block is generated based on the motion information of the adjacent block or whether the motion information is encoded / decoded.

Note that the motion information candidates used for block merging of the prediction block T2 without dividing the prediction block T2 are the motion information of the block Ba and the motion information of the block Bb, and the motion information of the block Ba and the motion of the block Bb One of the information may be selected as motion information used for generating the prediction signal of T2, but in that case, the flow of FIG. 7 needs to be extended as follows.
1. After Step S158 in FIG. 4 and Step S259 in FIG. 7, a flow of “obtaining the prediction block division type of the adjacent block and deriving the number of blocks adjacent to the prediction block” is added.
2. Step S159 in FIG. 4 and step S260 in FIG. 7 are changed to “two or more pieces of selection candidate motion information”.
3. Steps for encoding / decoding block selection information are added after step S159 in FIG. 4 and step S260 in FIG.

  Hereinafter, the flow of FIG. 13 will be described. In FIG. 13, first, in step S <b> 261 a, the prediction block division type of the encoded block that contacts the target prediction block is acquired. Next, in step S261b, the number M of prediction blocks adjacent to the block boundary adjacent to the adjacent block that does not belong to the target coding block is derived from the acquired prediction block division type. For example, in the case shown in FIG. 12B, M = 2. In step S261c, it is determined whether or not the value of M is greater than 1 (M> 1). When M> 1, the target prediction block is divided into M virtual blocks, and the motion information of M adjacent blocks is set in the divided M virtual blocks (additionally merged for each virtual block) The identification information may be sent to determine whether or not to decode the motion information). When M = 1, the motion information of the adjacent block that is a candidate for block merging is set as the motion information of the target prediction block.

As described above, according to FIGS. 12 and 13, the selection of motion information candidates in the example of FIG. 11 is performed based on the following information.
1) Number of encoded / decoded predicted blocks in target coding block 2) Predicted block partition type of target coded block 3) Predicted block partition type of block adjacent to target predicted block In FIG. Although an example of division has been described, the same processing can be applied to an example of horizontal division (division into a plurality of blocks arranged in the vertical direction) as in blocks 306 and 307 in FIG.

  Further, the following modifications are possible.

  (Candidate for motion information)

  In the above description, the motion information of the block that is in contact with the upper side and the left side of the prediction block is used as a candidate for block merging. However, as shown in FIGS. 14A and 14B and FIG. It is also possible to apply a restriction based on the prediction block division type of the generalized block and adjacent blocks. 14A and 14B show block merging candidates using motion information of adjacent blocks on the side where two or more adjacent blocks are in contact between the upper side and the left side of the prediction block when two adjacent blocks exist. An example to be excluded from is shown. In this case, encoding of merge block selection information becomes unnecessary, and additional information can be reduced. Motion information candidates used for block merging of the prediction block T1 of FIG. 14A and the prediction block T1 of FIG. 14B are determined as the motion information of block B and block A, respectively.

  FIG. 15A shows a method of automatically selecting motion information candidates used for block merging of the prediction blocks T1 and T2 according to the prediction block division type of the target coding block.

  FIG. 15B shows an example in which the prediction block to which block merging is applied is limited according to the prediction block division type of the target coding block and the number of encoded / decoded blocks in the target coding block. Yes. In the example shown in FIG. 3, when the motion information of the block T1 and the block D match, the motion information of the block D is excluded from the motion information candidates used for the block merging of the block T2. In the example shown in (2), without comparing the motion information of the block T1 and the block D, the block D is excluded from the block merging candidates based on the number of encoded / decoded predicted blocks in the target encoded block. Thus, the prediction block to which block merging is applied may be limited by the number of motion vectors encoded in the target encoding block.

  Furthermore, it is also possible to apply a restriction depending on the block size of two adjacent blocks in contact with the upper left corner of the prediction block and the block size of the prediction block. For example, when the size of the right side of an adjacent block that touches the left side of the target prediction block is smaller than a preset size (for example, half or 1/4 of the width of the left side of the prediction block), the motion information of the adjacent block Can be excluded from the block merging candidates of the target prediction block.

  In this way, by restricting motion information candidates, the code amount of block merging information can be reduced.

  (Selection of motion information candidates)

Selection of motion information candidates is performed based on the following information, but the method of use is not limited to the method described above. Means for selecting motion information candidates using these pieces of information can be implemented by the configuration shown in FIGS.
1) Number of encoded / decoded prediction blocks in the target coding block 2) Prediction block division type of the target coding block 3) Prediction block division type of a block adjacent to the target prediction block 4) In the target coding block 5) Motion information and prediction mode of the block adjacent to the target prediction block (intra prediction / inter prediction)

  (Coding of prediction block)

  In the above description, encoding / decoding of the prediction block in the encoding block is performed in the raster scan order, but selection of motion information candidates used for the block merging described above is performed in an arbitrary order. The present invention is also applicable when a prediction block is encoded / decoded. For example, in the example of FIG. 3, when the prediction block T2 of the target encoding block 400 is encoded / decoded first, the motion vector of the prediction block T2 is used as a candidate of motion information used for block merging of the prediction block T1. Is not included.

  (Block shape)

  In the above description, the partial area in the coding block is always rectangular, but may be any shape. In this case, shape information may be included in the prediction information of the encoded block.

  (Converter, Inverter)

  The residual signal conversion process may be performed with a fixed block size, or the target area may be subdivided according to the partial area and the conversion process may be performed.

  (Forecast information)

  In the above description, the prediction signal generation method is described as inter-screen prediction (prediction using a motion vector and reference screen information), but the prediction signal generation method is not limited to this. The prediction signal generation processing described above can also be applied to prediction methods including intra prediction and luminance compensation. In this case, mode information, luminance compensation parameters, and the like are included in the prediction information.

  In FIG. 10, the intra-screen prediction block in the adjacent block is virtually integrated into the inter-screen prediction block, but the prediction block is virtually divided regardless of the prediction mode of the adjacent block, and the prediction block is determined by the intra-screen prediction. The partial signal may be predicted.

  (Color signal)

  In the above description, the color format is not particularly described, but the prediction signal generation processing may be performed separately from the luminance signal for the color signal or the color difference signal. Further, the prediction signal generation processing may be performed in conjunction with the luminance signal processing.

  (Block noise removal processing)

  Although not described in the above description, when the block noise removal process is performed on the reproduced image, the noise removal process may be performed on the boundary portion of the partial region. When the prediction block is virtually divided in the examples shown in FIGS. 9, 10, and 12, the block noise removal process may be applied to the boundaries of the virtually divided blocks.

  The image predictive encoding method and the image predictive decoding method according to the present embodiment can be provided by being stored in a recording medium as a program. Examples of the recording medium include a recording medium such as a floppy disk (registered trademark), CD-ROM, DVD, or ROM, or a semiconductor memory.

  FIG. 16 is a block diagram illustrating modules of a program that can execute the image predictive coding method. The image predictive coding program P100 includes a block division module P101, a motion information estimation module P102, a prediction signal generation module P103, a storage module P104, a subtraction module P105, a transform module P106, a quantization module P107, an inverse quantization module P108, and an inverse transform. A module P109, an addition module P110, a quantized transform coefficient encoding module P111, a prediction division type selection module P112, a prediction information storage module P113, and a prediction information encoding module P114 are provided. The functions realized by the above modules being executed by a computer are the same as the functions of the image predictive coding apparatus 100 described above. That is, block division module P101, motion information estimation module P102, prediction signal generation module P103, storage module P104, subtraction module P105, transform module P106, quantization module P107, inverse quantization module P108, inverse transform module P109, addition module P110 , Quantized transform coefficient encoding module P111, prediction division type selection module P112, prediction information storage module P113, prediction information encoding module P114, block divider 102, motion information estimator 114, prediction signal generator 103, frame memory 104, subtractor 105, converter 106, quantizer 107, inverse quantizer 108, inverse transformer 109, adder 110, quantized transform coefficient encoder 111, prediction block division type selector 13, the prediction information memory 115, respectively to perform the same functions in a computer and the prediction information encoder 116.

  FIG. 17 is a block diagram showing modules of a program that can execute the image predictive decoding method. The image predictive decoding program P200 includes a quantized transform coefficient decoding module P201, a prediction information decoding module P202, a prediction information storage module P113, an inverse quantization module P206, an inverse transform module P207, an addition module P208, a prediction signal generation module P103, and a storage module. P104 is provided.

  The functions realized by executing the modules are the same as those of the components of the image predictive decoding apparatus 200 described above. That is, the quantization transform coefficient decoding module P201, the prediction information decoding module P202, the prediction information storage module P113, the inverse quantization module P206, the inverse transformation module P207, the addition module P208, the prediction signal generation module P103, and the storage module P104 are quantized. The transform coefficient decoder 207, the prediction information decoder 208, the prediction information memory 115, the inverse quantizer 203, the inverse transformer 204, the adder 205, the prediction signal generator 103, and the frame memory 104 have the same functions. To run.

  The image predictive encoding program P100 or the image predictive decoding program P200 configured as described above is stored in the recording medium SM and executed by a computer described later.

  FIG. 18 is a diagram illustrating a hardware configuration of a computer for executing a program recorded in a recording medium, and FIG. 19 is a perspective view of the computer for executing a program stored in the recording medium. Note that a program that executes a program stored in a recording medium is not limited to a computer, and may be a DVD player, a set-top box, a mobile phone, or the like that includes a CPU and performs processing and control by software.

  As shown in FIG. 19, a computer C10 includes a reading device C12 such as a floppy disk drive device, a CD-ROM drive device, a DVD drive device, a working memory (RAM) C14 in which an operating system is resident, and a recording medium SM. A memory C16 for storing the program stored in the memory, a display device C18 such as a display, a mouse C20 and a keyboard C22 as input devices, a communication device C24 for transmitting and receiving data and the like, and a CPU for controlling execution of the program C26. When the recording medium SM is inserted into the reading device C12, the computer C10 can access the image predictive encoding / decoding program stored in the recording medium SM from the reading device C12, and according to the image encoding or decoding program, It becomes possible to operate as an image encoding device or an image decoding device according to the present embodiment.

  As shown in FIG. 18, the image predictive encoding program and the image decoding program may be provided via a network as a computer data signal CW superimposed on a carrier wave. In this case, the computer C10 can store the image predictive coding program or the image decoding program received by the communication device C24 in the memory C16, and can execute the image predictive coding program or the image predictive decoding program.

  Hereinafter, another embodiment will be described. FIG. 24 is a diagram schematically illustrating a configuration of a video encoding device according to an embodiment. The moving picture encoding apparatus 10 shown in FIG. 24 includes a block divider 501, a small section generator 502, a frame memory 503, a motion detector 504, a prediction signal generator 505, a motion predictor 506, a subtractor 507, a residual signal. A generator 508, a converter 509, a quantizer 510, an inverse quantizer 511, an inverse transformer 512, an adder 513, and an entropy encoder 514 are provided. An input video signal (moving image signal) input to the moving image encoding apparatus 10 is composed of a time series of image signals (hereinafter referred to as frame image signals) in units of frames.

  The block divider 501 sequentially selects a frame image signal to be encoded, that is, an input image, from the input video signal input via the line L501. The block divider 501 divides the input image into a plurality of sections, that is, blocks. The block divider 501 sequentially selects a plurality of blocks as target blocks for encoding processing, and outputs pixel signals of the target blocks (hereinafter referred to as target block signals) via a line L502.

  In the moving image encoding device 10, the following encoding processing is performed in units of blocks. Note that the block divider 501 can divide an input image into a plurality of blocks of 8 × 8 pixels, for example. However, the size and shape of the block may be arbitrary. The block may be, for example, a 32 × 16 pixel block or a 16 × 64 pixel block.

The small section generator 502 divides the target block input via the line L502 into a plurality of small sections. FIG. 25 is a diagram for explaining generation of a small section. As shown in FIG. 25, the small section generator 502 divides the target block P into two small sections SP1 and SP2 by a straight line Ln described by the linear expression of Expression (1).
y = mx + k (1)

  For example, the small block generator 502 obtains the prediction signal of the small block SP1 and the prediction signal of the small block SP2 while changing the parameters m and k, and between the prediction signal of the small block SP1 and the image signal of the small block SP1. And m and k that minimize the error between the prediction signal of the small section SP2 and the image signal of the small section SP2 can be determined as parameters of the straight line Ln.

  The small section generator 502 determines the shape information for specifying the shape of the small section in the target block P, that is, the shape information for specifying the shapes of the first small section SP1 and the second small section SP2. The parameters m and k in the equation (1) are output via the line L504.

The linear expression representing the straight line Ln may be arbitrary. For example, the straight line Ln may be expressed as shown in Expression (2).
y = −x / tan θ + ρ / sin θ (2)
In this case, the shape information is θ and ρ.

  Further, the shape information may be information indicating any two points along which the straight line Ln passes, for example, the intersection of the straight line and the boundary of the block P. Further, the blocks do not necessarily have to be separated by straight lines, and small sections may be generated based on a pattern selected from a plurality of patterns prepared in advance. In this case, information such as an index for specifying the selected pattern can be used as the shape information.

  In the following description, coordinates are set so that the upper left position of the target block is the origin, and the small section including the upper left pixel in the target block P is referred to as a first small section, and the other is referred to as a second small section. However, an arbitrary method, for example, a small section that does not include the center position in the target block may be the first small section, and the other may be the second small section. In this case, the shape information may be block boundary intersection information or pattern identification information.

  The frame memory 503 stores a previously reproduced image signal input via the line L505, that is, a frame image signal that has been encoded in the past (hereinafter referred to as a reference frame image signal). The frame memory 503 outputs a reference frame image signal via the line L506.

  The motion detector 504 receives a target block signal input via the line L502, block shape information input via the line L504, and a reference frame image signal input via the line L506. The motion detector 504 searches for a signal similar to the image signal of the small section to be processed from image signals within a predetermined range of the reference frame image signal, and calculates a motion vector. This motion vector is a spatial displacement amount between a region in the reference frame image signal having a pixel signal similar to the image signal of the small section to be processed and the target block. The motion detector 504 outputs the calculated motion vector via the line L507.

  Note that the motion detector 504 may simultaneously detect a motion vector for the target block and determine whether to generate a prediction signal by dividing the target block into two small sections. In this determination, the error between the prediction signal of the target block and the image signal of the target block is the error between the prediction signal generated by dividing the target block into two small sections and the image signal of the two small sections. If it is smaller, it may be determined not to divide the target block into small sections. When making such a determination, the information indicating the result of the determination is encoded as the partitioning permission / inhibition information, and the shape information can be encoded only when the partitioning permission / inhibition information indicates that the target block is divided into small sections. .

  The prediction signal generator 505 is based on the motion vector input via the line L507 and the block shape information input via the line L504 from the image signal within the predetermined range of the reference frame image signal input via the line L506. Thus, a prediction signal of the image signal of the small section to be processed is generated.

  The prediction signal generator 505 generates a prediction signal of the target block by synthesizing the prediction signals of the small sections in the target block. The prediction signal generator 505 outputs the generated prediction signal via the line L508. Note that the prediction signal may be generated by intra-screen prediction instead of inter-screen prediction.

  The motion predictor 506 has processed the shape information of the block input via the line L504, the motion vector input via the line L507, and the processed block that is the previous block or small block in order from the small block to be processed. Based on the motion vector of the partial area, a predicted motion vector of the processing target small section in the target block is generated. The motion predictor 506 outputs the generated predicted motion vector via the line L509.

  The motion predictor 506 may select one predicted motion vector from a plurality of predicted motion vector candidates. In this case, the motion predictor 506 also outputs instruction information for specifying one selected predicted motion vector via the line L510. Note that the output of the instruction information can be omitted by narrowing down the predicted motion vector candidates of the processing target sub-partition to one according to a predetermined rule shared with the decoding side.

  The subtractor 507 subtracts the predicted motion vector input via the line L509 from the motion vector of the processing target small block input via the line L507 to generate a differential motion vector. The subtractor 507 outputs the generated differential motion vector via the line L511.

  The residual signal generator 508 generates a residual signal by subtracting the prediction signal of the target block input via the line L508 from the target block signal input via the line L502. Residual signal generator 508 outputs the generated residual signal via line L512.

  The converter 509 generates a transform coefficient by performing orthogonal transform on the residual signal input via the line L512. The converter 509 outputs the generated conversion coefficient via the line L513. For example, DCT can be used for this orthogonal transformation. However, any conversion can be used for the conversion used by the converter 509.

  The quantizer 510 generates a quantized transform coefficient by quantizing the transform coefficient input via the line L513. The quantizer 510 outputs the generated quantized transform coefficient via the line L514.

  The inverse quantizer 511 generates an inverse quantization transform coefficient by inversely quantizing the quantization transform coefficient input via the line L514. The inverse quantizer 511 outputs the generated inverse quantization transform coefficient via the line L515.

  The inverse transformer 512 generates a reproduction residual signal by performing inverse orthogonal transform on the inverse quantization transform coefficient input via the line L515. The inverse converter 512 outputs the generated reproduction residual signal via the line L516. The inverse transformation used by the inverse transformer 512 is a process that is symmetric to the transformation of the converter 509.

  Note that the conversion is not essential, and the moving image encoding apparatus may not include the converter 509 and the inverse converter 512. Similarly, quantization is not essential, and the moving picture coding apparatus may not include the quantizer 510 and the inverse quantizer 511.

  The adder 513 generates a reproduction image signal by adding the reproduction residual signal input via the line L516 and the prediction signal of the target block input via the line L508. The adder 513 outputs the reproduced image signal as an already reproduced image signal via the line L505.

  The entropy encoder 514 includes a quantized transform coefficient input via the line L514, target block shape information input via the line L504, prediction motion vector indication information input via the line L510, The differential motion vector input via L511 is encoded. The entropy encoder 514 generates a compressed stream by multiplexing the codes generated by encoding, and outputs the compressed stream to the line L517.

  The entropy encoder 514 can use an arbitrary encoding method such as arithmetic encoding or run-length encoding. Also, the entropy encoder 514 generates occurrence probability when arithmetically encoding the instruction information of the predicted motion vector input via the line L510 based on the shape information of the target block input via the line L504. Can be determined adaptively. For example, the entropy encoder 514 can set the occurrence probability of the instruction information indicating the motion vector of the partial area in contact with the processing target small section to be high.

  FIG. 26 is a diagram illustrating a configuration of a motion predictor according to an embodiment. As illustrated in FIG. 26, the motion predictor 506 includes a motion vector memory 5061, a motion reference destination candidate generator 5062, and a predicted motion vector generator 5063.

  The motion vector memory 5061 stores the motion vector of the processed partial area, and outputs the encoded motion vector via the line L5061 in order to derive the predicted motion vector of the processing target sub-partition.

  Based on the shape information input via the line L504, the motion reference destination candidate generator 5062 generates a predicted motion vector candidate from the motion vector of the partial region input via the line L5061 based on the method described later. The motion reference destination candidate generator 5062 outputs the generated predicted motion vector candidate via the line L5062.

  The motion vector predictor generator 5063 selects a candidate having the smallest difference from the motion vector of the small block to be processed from the motion vector predictor candidates input via the line L5062. The motion vector predictor generator 5063 outputs the selected candidate as a motion vector predictor via the line L509. In addition, instruction information for specifying the selected candidate is output via the line L510.

  Note that the output of the instruction information can be omitted by limiting the number of candidates generated in the motion reference destination candidate generator to one. Although the method of limiting the number of candidates to one is not limited, for example, a method using an intermediate value of three candidates, a method using an average value of two candidates, or selecting one from a plurality of candidates Any method can be used, such as a method for determining the priority order for the operation.

  Hereinafter, the operation of the moving image encoding apparatus 10 will be described, and a moving image encoding method according to an embodiment will be described. FIG. 27 is a flowchart of a video encoding method according to an embodiment.

  As shown in FIG. 27, in one embodiment, first, in step S501, the block divider 501 divides an input image into a plurality of blocks. In subsequent step S502, the small section generator 502 divides the target block into a plurality of small sections as described above. The small section generator 502 generates shape information as described above.

  Next, in step S503, as described above, the motion detector 504 obtains the motion vector of the small section to be processed. In subsequent step S504, as described above, the prediction signal generator 505 generates a prediction signal of the target block using the motion vector of each subsection in the target block and the reference frame image signal.

  Next, in step S505, the motion predictor 506 obtains a predicted motion vector. In addition, the motion predictor 506 generates instruction information for specifying a candidate selected from a plurality of prediction motion vector candidates. Details of the processing in step S505 will be described later. In the subsequent step S506, the subtracter 507 obtains a difference between the motion vector of each small block and the predicted motion vector as described above, and generates a differential motion vector.

  Next, in step S507, the residual signal generator 508 calculates a difference between the image signal of the target block and the prediction signal, and generates a residual signal. In subsequent step S508, converter 509 performs orthogonal transform on the residual signal to generate a transform coefficient. In subsequent step S509, the quantizer 510 quantizes the transform coefficient to generate a quantized transform coefficient. In subsequent step S510, the inverse quantizer 511 performs inverse quantization on the quantized transform coefficient to generate an inverse quantized transform coefficient. In subsequent step S511, the inverse transformer 512 inversely transforms the inverse quantization transform coefficient to generate a reproduction residual signal.

  Next, in step S512, the adder 513 generates a reproduced image signal by adding the prediction signal of the target block and the reproduction residual signal. In a subsequent step S513, the reproduced image signal is stored as an already reproduced image signal by the frame memory 503.

  Next, in step S514, the entropy encoder 514 encodes the quantized transform coefficient, the shape information of the target block, the instruction information of the predicted motion vector, and the difference motion vector.

  Next, in step S515, it is determined whether all blocks have been processed. If all the blocks have not been processed, the processing from step S502 is continued for unprocessed blocks. On the other hand, if all the blocks have been processed, the process ends.

  Hereinafter, the operation of the motion predictor 506 will be described in more detail. FIG. 28 is a flowchart illustrating processing of the motion predictor according to an embodiment. The motion predictor 506 outputs a prediction motion vector (hereinafter referred to as PMV) and instruction information for specifying the PMV according to the flowchart shown in FIG.

  As shown in FIG. 28, in the process of the motion predictor 506, first, in step S505-1, the value of the counter i is set to zero. Hereinafter, it is assumed that when i = 0, the process for the first small section is performed, and when i = 1, the process for the second small section is performed.

  Next, in step S505-2, PMV candidates of the processing target small section are generated from the motion vectors of the processed partial areas according to a method described later. The number of PMV candidates is two in this example. That is, as the PMV candidates, the motion vector of the processed partial area on the left side of the processing target sub-partition and the motion vector of the upper processed partial area are set as the predicted motion vector candidates of the processing target sub-partition. obtain. In step S505-2, the number of generated candidates is set to Ncand.

  Next, in step S505-3, it is determined whether NCand is “0”. If NCand is “0” (Yes), the process proceeds to step S505-4. If NCand is not “0” (No), the process proceeds to step S505-5.

  In step S505-4, PMV is set to a zero vector, and the process proceeds to step S505-10. At this time, the PMV may be set to a motion vector of a predetermined block, a motion vector of a partial area processed immediately before the processing target subsection, or the like, instead of the zero vector.

  In step S505-5, it is determined whether NCand is “1”. If NCand is “1” (Yes), the process proceeds to step S505-10. If NCand is not “1” (No), the process proceeds to step S505-6.

  In step S505-6, a PMV is selected from the PMV candidates generated in step S505-2. As the PMV, a candidate that minimizes the difference with respect to the motion vector of the small section to be processed can be selected.

  In step S505-7, it is determined whether the PMV selected in step S505-6 is a left candidate, that is, a motion vector of the left partial region. If the PMV selected in step S505-6 is a left candidate (Yes), the process proceeds to step S505-8. If the PMV selected in step S505-6 is not the left candidate (No), the process proceeds to step S505-9.

  In step S505-8, instruction information pmv_left_flag = 1 is output with the motion vector of the left partial region of the processing target subsection being PMV. On the other hand, in step S505-9, instruction information pmv_left_flag = 0, in which the motion vector of the upper partial area of the processing target small section is PMV, is output.

  Next, in step S505-10, the PMV remaining as a candidate is output. In subsequent step S505-11, "1" is added to the value of counter i.

  Next, in step S505-12, it is determined whether or not the value of the counter i is smaller than “2”. If the value of counter i is smaller than “2” (Yes), the process proceeds to step S505-2. If the value of the counter i is not smaller than “2” (No), the process is terminated.

  In step S505-2, by limiting the number of generated candidates to one, steps S505-5, S505-6, S505-7, S505-8, and S505-9 can be omitted. This limitation method is not limited. However, as described above in the description of the motion vector predictor generator 5063, for example, a method using an intermediate value of three candidates and a method using an average value of two candidates. A method of determining a priority order for selecting one candidate from a plurality of candidates can be used. In the form in which the number of candidates generated in step S505-2 is limited to one, if NCand is not “0” in step S505-3 (No), the process proceeds to step S505-10.

  Hereinafter, the method of generating the predicted motion vector candidates of the processing target small section in step S505-2 will be described in more detail. FIG. 29 is a diagram illustrating an example of a small section of a target block and a surrounding partial area.

  As illustrated in FIG. 29, the motion reference destination candidate generator 5062 refers to the partial area U1 and the partial area L1 with respect to the first small section SP1, and each partial area has been processed by inter-frame prediction. In this case, the motion vector of the partial area is set as a candidate for the predicted motion vector of the first small section SP1. Similarly, the motion reference destination candidate generator 5062 refers to the partial area U2 or the partial area L2 for the second small section, and generates a predicted motion vector candidate for the second small section. Here, the partial areas U1, L1, U2, and L2 are blocks or small sections around the target block P, and are areas that are units for generating a prediction signal. Further, the partial region may be a block (for example, generated by dividing into a single shape) prepared for generating a motion vector predictor candidate regardless of a unit for generating a prediction signal. .

  The partial area U1 is a partial area including the pixel Pi1 (0, −1) adjacent to the uppermost leftmost pixel F (0, 0) of the first small section SP1, and has been processed in contact with the small section SP1. It is a partial area. The partial area L1 is a partial area including the pixel Pi2 (-1, 0) adjacent on the left side to the upper left pixel F (0, 0) of the first small section SP1, and is in contact with the first small section SP1. It is a partial area. The partial region U2 is a partial region adjacent to the partial region including the pixel Pi3 (x1, −1) on the right side and is in contact with the x-axis. The partial region L2 is a partial region adjacent to the partial region including the pixel Pi4 (−1, y1) on the lower side and is in contact with the y-axis.

The x coordinate x1 of the pixel Pi3 and the y coordinate y1 of the pixel Pi4 can be calculated by Expressions (3) and (4).
x1 = ceil (−k / m) (3)
y1 = ceil (k) (4)
Expressions (3) and (4) are values obtained by substituting y = 0 and x = 0 into the linear expression (1) representing the extended line Ln of the boundary dividing the first small section SP1 and the second small section SP2, respectively. To which the ceil (z) function is applied. ceil (z) is called a ceiling function, and is a function for deriving a minimum integer equal to or greater than z with respect to the real number z.

  A floor function may be used instead of the ceil function. floor (z) is called a floor function, and is a function for deriving the maximum integer equal to or smaller than z with respect to the real number z.

Moreover, x1 and y1 may be calculated by the equations (5) and (6).
x1 = ceil ((− 1−k) / m) (5)
y1 = ceil (−m + k) (6)
Expressions (5) and (6) are obtained by applying the ceil (z) function to values obtained by substituting y = −1 and x = −1 into Expression (1), respectively.

Whether or not the partial areas U2 and L2 exist is determined as described below. The condition for the partial region U2 to exist is to be within the screen and to satisfy Expression (7). Moreover, the conditions for the partial region L2 to exist are to be within the screen and to satisfy Expression (8).
0 <x1 (7)
0 <y1 (8)

  When the condition of Expression (7) is not satisfied, the partial region L2 exists between the second small section SP2 and the partial region U2. In that case, it is less likely that the partial area U2 farther from the second small section SP2 has a motion vector closer to the motion vector of the second small section SP2 than the partial area L2 near the second small section SP2. In such a case, the motion vector of the partial region U2 can be excluded from the prediction motion vector candidates according to the condition of Expression (7).

  Similarly, when the condition of Expression (8) is not satisfied, the partial region U2 exists between the second small section SP2 and the partial region L2. In that case, it is less likely that the partial area L2 far from the second small section SP2 has a motion vector closer to the motion vector of the second small section SP2 than the partial area U2 near the second small section SP2. In such a case, the motion vector of the partial region U2 can be excluded from the prediction motion vector candidates according to the condition of Expression (8).

In one example, conditions defined by the following expressions (9) and (10) may be used instead of the conditions of expressions (7) and (8).
0 <x1 <blocksizeX (9)
0 <y1 <blocksizeY (10)
Here, blocksizeX and blocksizeY are the number of pixels in the horizontal direction and the number of pixels in the vertical direction of the target block P, respectively. For example, when the target block P is an 8 × 8 pixel block, blocksizeX = 8 and blocksizeY = 8.

  By using the condition of Equation (9) or Equation (10), the motion vector of the partial region that is not in contact with the second small section SP2 out of the partial region U2 and the partial region L2 is excluded from the predicted motion vector candidates. Can do. Thereby, only prediction motion vector candidates that are considered to have high prediction accuracy can be left.

  By setting the partial areas U1, L1, U2, and L2 in this way, the predicted motion vector candidates of each sub-partition are the same as the processed partial areas on the same side with respect to the extension line of the boundary between the sub-partitions. Generated from motion vectors.

  Note that the candidate for the predicted motion vector of the small section SP2 is in the same area as the small section SP2 with respect to the extension line Ln of the boundary between the small section SP2 and the other small sections of the target block including the small section SP2. As long as the motion vector is generated from the motion vector of the partial region, the method is not limited to the method for generating the predicted motion vector of the above-described embodiment. For example, the partial region U2 may be a partial region including the pixel Pi3, and the partial region L2 may be a partial region including the pixel Pi4.

  Alternatively, the fact that the entire partial area is in the same area as the small section SP2 with respect to the line Ln may be a condition for adding the motion vector of the partial area to the predicted motion vector candidate of the small section SP2. In this case, for example, a method of inspecting the positions of all the vertices of the partial area can be used.

Even if the partial area is not completely included in the area on the same side as the small section with respect to the extension line, the motion vector of the partial area may be used as a candidate for the predicted motion vector of the small section. FIG. 30 is a diagram illustrating another example of the small section of the target block and the surrounding partial area. As an example shown in FIG. 30, the motion vectors of the partial regions R _A , R _B , R _G , and R _E may be used as the predicted motion vector candidates of the first small section SP1. Further, the candidate predicted motion vector of the second subdivision SP2, it may be added to the predicted motion vector of the partial region R _E.

  In the description related to FIG. 28 and FIG. 29, the maximum number of motion vectors that are candidates for the predicted motion vector is two, but two may be selected from the motion vectors obtained under any of the above-described conditions. For example, the motion vector of the partial region U2 shown in FIG. 29 and the motion vector of the partial region adjacent to the partial region U2 may be set as the predicted motion vector candidates. Similarly, the motion vector of the partial area L2 and the motion vector of the partial area adjacent to the partial area U2 may be candidates for the predicted motion vector. Furthermore, three or more motion vectors from the motion vectors specified by any of the above-described conditions may be set as predicted motion vector candidates. Further, an average value or median value of a plurality of motion vector predictor candidates may be added to the motion vector predictor candidates.

  Further, as a method for limiting the number of motion vector predictor candidates generated in step S505-2 in FIG. 28 to a maximum of one, it is possible to use section shape information. For example, a motion vector of a partial area in which the length of the part in contact with the small section among the encoded partial areas in contact with the small section to be processed is the maximum may be used as a predicted motion vector candidate. Also, the motion vector of the encoded partial region with the shortest shortest distance from the processing target sub-partition may be used as the predicted motion vector candidate of the sub-partition.

  Further, the above-described method for generating motion vector predictor candidates can be applied to a small section having an arbitrary shape. FIG. 31 is a diagram showing still another example of the small section of the target block and the surrounding partial area. FIG. 31A shows a small section defined by a line Ln having a different coordinate and inclination from the line Ln shown in FIG. FIG. 31B shows a small section defined by a line Ln whose inclination is substantially symmetric with respect to the y-axis with respect to the line Ln shown in FIG. 29 and whose coordinates intersecting the y-axis are different. FIG. 31C shows a small section defined by two lines Ln1 and Ln2. FIG. 31D shows a small section defined by two lines Ln1 and Ln2 that intersect each other. Even if the boundary extension lines as shown in FIGS. 31A to 31D are used as a reference, the partial motion regions L2 and U2 having motion vectors that can be candidates for the motion vector predictor of the small section SP2 are converted into the motion vector predictors described above. It is possible to specify by the method of generating candidates.

  Further, the small section is not limited to only the small section divided by a straight line. For example, even when selecting the shape of a small section from a predetermined pattern, the encoded partial area belonging to the same side area as the processing target small section with respect to the extension line of the boundary between the small sections Can be used as candidates for the predicted motion vector. In addition, when the pattern of the shape of the small section is determined in advance, it is also possible to determine in advance a partial region having a motion vector as a candidate for a predicted motion vector for each shape pattern. This pattern may include a pattern for dividing the target block into rectangles.

  Further, the above-described prediction motion vector selection method can also be applied as a motion vector selection method when generating a prediction signal of a small block to be processed using a motion vector of an encoded partial region. That is, the prediction signal of the small section to be processed may be generated using the prediction motion vector selected in step S505-2 in FIG. In this case, since it is not necessary to encode the differential motion vector, the predicted motion vector output from the motion predictor 506 is output to the predicted signal generator 505 instead of the subtracter 507.

  Furthermore, the moving image encoding device 10 may determine whether or not to encode the differential motion vector, and may encode application information that specifies the determination result. In this variation, the motion predictor 506 may include a function of switching whether to output the predicted motion vector to the subtracter 507 or to the predicted signal generator 505 based on the application information.

  In this modification, it is not preferable that the motion vectors of all the small sections in the target block are the same because it makes no sense to divide the target block. In other words, when generating motion vector candidates for the small section to be processed in step S505-2 in FIG. 28, the motion vectors of the encoded small sections included in the target block may be excluded from the candidates. For example, when the target block is divided into two small sections and the motion vector of the first small section is encoded first, the motion vector of the first small section is the predicted motion vector of the second small section. Excluded from the candidate. Further, when the motion vector of the first small section and the motion vector of the partial area U2 are the same, the motion vector of the partial area U2 may not be used for generating the predicted motion vector of the second small section.

  When instructing whether or not to encode the differential motion vector, the occurrence probability when the application information is arithmetically encoded may be adaptively determined according to the shape information of the small sections. For example, the occurrence probability for application information indicating that the differential motion vector of the first small section is not encoded is set higher than the occurrence probability for application information indicating that the differential motion vector of the second small section is not encoded. Can do. This is because the second sub-partition may not touch the encoded partial area, while the first sub-partition always touches the encoded partial area. Thus, by setting the occurrence probability in this way, This is because the code amount of the application information can be reduced.

Here, for simplification, the effect of the embodiment will be described with reference to FIG. 32 showing an example in which the target block is divided into rectangles. In this example, the target block P is divided into a left small section SP1 and a right small section SP2 by a straight line Ln. In this example, the motion vector of the motion vector and the partial region R _B of the first small compartment SP1 becomes the candidate of the predicted motion vector of the second subsection SP2.

In the example shown in FIG. 32, if the prediction signal of the second small section SP2 is generated using the motion vector of the first small section SP1, the prediction signal of the first small section SP1 and the prediction signal of the second small section SP2 are the same. It will be generated by a motion vector, and it makes no sense to divide the target block into two small sections. Therefore, the prediction signal of the second subsection SP2 can be produced using a motion vector of an upper partial region R _B of the cubicle SP2. Therefore, the encoding device and the decoding device, in the example shown in FIG. 32, by preliminarily decided to generate a prediction signal of the second subdivision SP2 using the motion vector of the partial region R _B, predicted motion vector The number of candidates decreases, and it is no longer necessary to send instruction information for indicating one prediction motion vector from a plurality of prediction motion vector candidates.

Further, the moving image coding apparatus 10 determines whether or not the differential motion vector can be encoded (the motion predictor 506 outputs the predicted motion vector to the subtractor 507 based on the application information, or the predicted signal generator 505 Consider switching the output. At this time, the motion vector of the partial region R _B is given the same as the motion vector of the first sub-compartment SP1, the predicted motion vector of the two predicted motion second subsection also select one of the candidate vectors SP2 first It becomes the same as the motion vector of one small section SP1. Accordingly, when the two prediction motion vector candidates are the same, the encoding device and the decoding device generate a prediction signal for the second sub-partition SP2 in advance using a motion vector obtained by adding the difference motion vector and the prediction motion vector. By deciding, it is not necessary to send application information instructing whether or not to encode the differential motion vector in addition to the instruction information.

As shown in FIG. 33, when the target block is divided into three or more small sections, the first small section SP1, the second small section SP2, and the third small section SP3 have the same motion vector. And if only the fourth small section SP4 has a different motion vector, it is meaningful to divide the target block. Therefore, in this case, the prediction signal and the prediction signal of the third subsection SP3 of the second subsection SP2, respectively partial region R _B, rather than the motion vector of the partial region R _E, a motion vector of the first subsection SP1 You may generate using. However, for the fourth sub-partition SP4, if the motion vector of the second sub-partition SP2 and the motion vector of the third sub-partition SP3 are the same, the two motion vector candidates are the same. By determining the rule in the decoding device, there is no need to send instruction information for instructing one prediction motion vector. Furthermore, when the first small section SP1, the second small section SP2, and the third small section SP3 have the same motion vector, the prediction signal of the fourth small section SP4 is sent to the second small section SP2 or the third small section SP3. If the motion vector is used for generation, all four sections have the same motion vector. Therefore, by defining rules in advance in the encoding device and the decoding device, in addition to the instruction information, the difference motion vector It is no longer necessary to send application information indicating whether or not to encode.

  Hereinafter, a moving picture decoding apparatus according to an embodiment will be described. FIG. 34 is a diagram schematically illustrating a configuration of a video decoding device according to an embodiment. The video decoding device 20 shown in FIG. 34 is a device that can generate a video by decoding the compressed stream generated by the video encoding device 10.

  As shown in FIG. 34, the moving picture decoding apparatus 20 includes a data decoder 601, a motion predictor 602, an adder 603, an inverse quantizer 604, an inverse transformer 605, a frame memory 606, a prediction signal generator 607, and , An adder 608 is provided.

  The data decoder 601 analyzes the compressed stream input via the line L601. The data decoder 601 sequentially performs the following processing on a block to be decoded (hereinafter referred to as a target block).

  The data decoder 601 decodes the encoded data related to the target block in the compressed stream, restores the quantized transform coefficient of the target block, and outputs the quantized transform coefficient via the line L602. Further, the data decoder 601 decodes the encoded data, restores the shape information of the target block, and outputs the shape information via the line L603. At this time, if the blockability information indicating whether or not to divide the target block is restored, and the blockability information indicates that the block is not divided, the shape information may not be restored. .

  Further, the data decoder 601 restores instruction information for each subsection in the target block, that is, information indicating one of a plurality of motion vector predictor candidates, by decoding the encoded data, The instruction information is output via line L604. Also, the data decoder 601 restores the differential motion vector of the target block by decoding the encoded data, and outputs the differential motion vector via the line L605. In addition, the data decoder 601 can adaptively determine the occurrence probability in decoding the encoded data when restoring the motion vector instruction information based on the shape information of the target block. As this method, for example, it is possible to increase the occurrence probability of the instruction information that indicates the motion vector of the partial area that is in contact with the processing target small section as the predicted motion vector.

  The motion predictor 602 performs processing based on the shape information input via the line L603, the motion vector of the previous partial region in the processing order input via the line L606, and the instruction information input via the line L604. Are generated, and the predicted motion vector is output via line L607. Note that the input of the instruction information can be omitted by narrowing the prediction motion vector candidates to one by a predetermined method.

  The adder 603 adds the predicted motion vector input via the line L607 and the differential motion vector input via the line L605 to generate a motion vector of the target block or a small section in the target block, and The vector is output via line L606.

  The inverse quantizer 604 generates an inverse quantized transform coefficient by inversely quantizing the quantized transform coefficient input via the line L602. The inverse quantizer 604 outputs the generated inverse quantization transform coefficient via the line L608.

  The inverse transformer 605 generates a reproduction residual signal by performing inverse orthogonal transform on the inverse quantization transform coefficient input via the line L608. The inverse converter 605 outputs the generated reproduction residual signal via the line L609.

  Note that when the generated reproduction residual signal is not quantized, the moving picture decoding apparatus 20 may not include the inverse quantizer 604. Similarly, when the generated reproduction residual signal is not converted, the video decoding device 20 may not include the inverse converter 605.

  The frame memory 606 stores an already reproduced image signal input via the line L610, that is, a frame image signal (hereinafter referred to as a reference frame image signal) preceding the processing target input image in processing order. The frame memory 606 outputs a reference frame image signal via the line L611.

  The prediction signal generator 607 is based on the motion vector input via the line L606 and the shape information input via the line L603 from the image signal within the predetermined range of the reference frame image signal input via the line L611. A prediction signal of an image of each small section in the target block is generated. The prediction signal generator 607 outputs the generated prediction signal via the line L612. In addition, although description is abbreviate | omitted in this specification, a prediction signal may be produced | generated by the prediction in a screen other than the prediction between screens.

  The adder 608 generates a reproduced image signal by adding the reproduction residual signal input via the line L609 and the prediction signal of the target block input via the line L612. The adder 608 outputs the reproduced image signal via the line L610.

  FIG. 35 is a diagram illustrating a configuration of a motion predictor according to an embodiment. As illustrated in FIG. 35, the motion predictor 602 includes a motion vector memory 6021, a motion reference destination candidate generator 6022, and a predicted motion vector generator 6023.

  The motion vector memory 6021 stores a motion vector input via the line L606. The motion vector stored in the motion vector memory 6021 is a motion vector of a partial area that has been processed before the target block or processing target sub-partition in the processing order. The motion vector memory 6021 outputs the stored motion vector via a line L6021, in order to derive a predicted motion vector for the processing target small section.

  The motion reference destination candidate generator 6022 generates a predicted motion vector candidate from the motion vector input via the line L6021, based on the shape information input via the line L603, by a method described later, and sets the line L6022 as the line L6022. Output via.

  The predicted motion vector generator 6023 determines a predicted motion vector based on the predicted motion vector instruction information input via the line L604 from the predicted motion vector candidates input via the line L6022, and the determined predicted motion The vector is output via line L607. Note that by limiting the number of candidates generated by the motion reference destination candidate generator to one, the input of instruction information for specifying candidates to be selected can be omitted.

  Hereinafter, an operation of the video decoding device 20 and a video decoding method according to an embodiment will be described. FIG. 36 is a flowchart of a video decoding method according to an embodiment. As shown in FIG. 36, in one embodiment, first, in step S621, as described above, the data decoder 601 decodes the encoded data in the compressed data for the target block, and the quantum of the target block. The conversion factor, shape information, and differential motion vector are restored. Further, in step S621, the partition availability information and the instruction information can be restored. Further, in step S621, the inverse quantizer 604 can generate an inverse quantized transform coefficient from the restored quantized transform coefficient, and the inverse transformer 605 can generate a reproduction residual signal from the inverse quantized transform coefficient. .

  Next, in step S622, the motion predictor 602 sets each small partition in the target block as a processing target, and obtains a predicted motion vector of the processing target small partition. In subsequent step S623, the adder 603 generates a motion vector by adding the predicted motion vector and the difference motion vector of the processing target sub-partition.

  Next, in step S624, the prediction signal generator 607 generates a prediction signal from the reference frame image signal in the frame memory 606 using the motion vector of the target block. In the subsequent step S625, the adder 608 adds the prediction signal of the target block and the reproduction residual signal to generate a reproduction image signal.

  Next, in step S626, the reproduced image signal generated in step S625 is stored in the frame memory 606 as an already reproduced image signal. In a succeeding step S627, it is determined whether or not processing has been performed on all blocks. If the processing for all the blocks has not been completed, the processing from step S621 is continued with the unprocessed block as the target block. On the other hand, if all the blocks have been processed, the process ends.

  Hereinafter, the operation of the motion predictor 602 will be described in detail. FIG. 37 is a flowchart illustrating processing of the motion predictor according to an embodiment. The motion predictor 602 generates a predicted motion vector according to the flowchart shown in FIG.

  In one embodiment, the value of the counter i is set to “0” in step S615-1. Hereinafter, it is assumed that when i = 0, the process for the first small section is performed, and when i = 1, the process for the second small section is performed.

  Next, in step S615-2, two candidate candidates (left candidate and upper candidate) that can be predicted motion vectors of the sub-parts to be processed among the motion vectors of the partial areas preceding in the processing order from the sub-parts to be processed. Candidates) are obtained according to the method described above with reference to FIGS. 29, 30, 31, 32, and 33.

  Next, in step S615-3, it is determined whether or not the number NCand of candidates generated in step S615-2 is “0”. If NCand is “0” (Yes), the process proceeds to step S615-4. If NCand is not “0” (No), the process proceeds to step S615-5.

  In step S615-4, the predicted motion vector PMV is set to a zero vector, and the process proceeds to step S615-11. At this time, instead of the zero vector, the motion vector of the block determined in advance or the motion vector of the partial area immediately before in the processing order with respect to the small section to be processed may be set as the predicted motion vector PMV.

  In step S615-5, it is determined whether the number NCand of candidates generated in step S615-2 is “1”. If NCand is “1” (Yes), the process proceeds to step S615-6. If NCand is not “1” (No), the process proceeds to step S615-7.

  In step S615-6, one candidate generated in step S615-2 is set as the PMV. Then, the process proceeds to step S615-11.

  In step S615-7, information pmv_left_flag for indicating the PMV to be selected is obtained from the candidates generated in step S615-2. Then, the process proceeds to step S615-8.

  In step S615-8, it is determined whether the value of pmv_left_flag is “1”. If the value of pmv_left_flag is “1” (Yes), the process proceeds to step S615-9. If the value of pmv_left_flag is not “1” (No), the process proceeds to step S615-10.

  In step S615-9, the motion vector of the partial region on the left side of the processing target subsection is set to PMV. Then, the process proceeds to step S615-11.

  In step S615-10, the motion vector of the partial region on the left side of the processing target subsection is set to PMV. Then, the process proceeds to step S615-11.

  In step S615-11, the set PMV is output. Then, the process proceeds to step S615-12.

  Next, in step S615-12, “1” is added to the value of the counter i. Then, the process proceeds to step S615-13.

  Next, in step S615-13, it is determined whether or not the value of the counter i is smaller than “2”. If the value of the counter i is smaller than “2” (Yes), the process proceeds to step S615-2. On the other hand, when the value of the counter i is not smaller than 2 (No), the process ends.

  In step S615-2, by limiting the number of prediction motion vector candidates to be generated to one, steps S615-5, S615-6, S615-7, S615-8, S615-9, S615-10 are performed. This process can be omitted. The method for this limitation is not limited as described above for the motion vector predictor generator 6023. For example, a method using an intermediate value of three candidates, a method using an average value of two candidates, A method of determining a priority order for selecting one predicted motion vector from the predicted motion vector candidates can be used. In this case, if NCand is not “0” in step S615-03 (No), the process proceeds to step S615-6.

  Further, the above-described method can be applied as a motion vector selection method in the case of generating a prediction signal of a processing target small section using a decoded motion vector. That is, the prediction signal of the small section to be processed may be generated using the prediction motion vector selected in step S615-2 in FIG. In this case, since it is not necessary to decode the differential motion vector, the predicted motion vector output from the motion predictor 602 is output to the predicted signal generator 607 instead of the adder 603.

  Further, the application information specifying whether or not to decode the differential motion vector may be decoded by the data decoder 601. In this variation, the motion predictor 602 may include a function of switching whether to output the predicted motion vector to the adder 603 or the predicted signal generator 607 based on the application information.

  In this modification, it is not preferable that the motion vectors of all the small sections in the target block are the same because it makes no sense to divide the target block. Therefore, in this modification, when generating the motion vector predictor candidate for the processing target sub-partition in step S615-2 in FIG. 37, the sub-parts included in the target block are processed in the processing order from the processing target sub-partition. The motion vector of the previous small section may be excluded from the prediction motion vector candidates. For example, when the target block is divided into two subsections and the motion vector of the first subsection is restored first, the motion vector of the first subsection is the predicted motion vector of the second subsection. Excluded from the candidate. Further, when the motion vector of the first small section and the motion vector of the partial area U2 are the same, the motion vector of the partial area U2 may not be used for generating the predicted motion vector of the second small section.

  Further, in this modified mode, the occurrence probability at the time of arithmetic decoding of the application information indicating whether or not to decode the differential motion vector can be adaptively determined according to the shape information. As this method, for example, the probability that the differential motion vector is not encoded is set higher in the first partition always in contact with the encoded region than in the second small partition that may not be in contact with the decoded partial region. obtain. Note that the effect of this modification is not described here because it has already been described with reference to FIGS.

  Hereinafter, a moving image encoding program for operating a computer as the moving image encoding device 10 and a moving image decoding program for operating the computer as a moving image decoding device 20 will be described.

  FIG. 38 is a diagram illustrating a configuration of a moving image encoding program according to an embodiment. FIG. 39 is a diagram illustrating a configuration of a moving image decoding program according to an embodiment. 38 and 39, FIG. 18 showing a hardware configuration of a computer according to an embodiment, and FIG. 19 which is a perspective view showing a computer according to an embodiment will be referred to.

  The moving image encoding program P10 shown in FIG. 38 can be provided by being stored in the recording medium SM. Also, the moving picture decoding program P20 shown in FIG. 38 can be stored and provided in the recording medium SM. The recording medium SM is exemplified by a floppy disk, a CD-ROM, a DVD, a ROM, or other recording medium, or a semiconductor memory.

  As described above, the computer C10 stores the reading device C12 such as a floppy disk drive device, a CD-ROM drive device, a DVD drive device, the working memory (RAM) C14 in which the operating system is resident, and the recording medium SM. A memory C16 for storing the recorded program, a display device C18 such as a display, a mouse C20 and a keyboard C22 as input devices, a communication device C24 for transmitting and receiving data and the like, and a CPU C26 for controlling execution of the program Can be provided.

  When the recording medium SM is inserted into the reading device C12, the computer C10 can access the moving image encoding program P10 stored in the recording medium SM from the reading device C12. 10 can be operated.

  Further, when the recording medium SM is inserted into the reading device C12, the computer C10 can access the moving image decoding program P20 stored in the recording medium SM from the reading device C12, and the moving image decoding device is obtained by the program P20. 20 can be operated.

  As shown in FIG. 19, the moving image encoding program P10 and the moving image decoding program P20 may be provided via a network as a computer data signal CW superimposed on a carrier wave. In this case, the computer C10 can store the moving image encoding program P10 or the moving image decoding program P20 received by the communication device C24 in the memory C16 and execute the program P10 or P20.

  As shown in FIG. 38, the moving image encoding program P10 includes a block division module M101, a small section generator module M102, a storage module M103, a motion detection module M104, a prediction signal generation module M105, a motion prediction module M106, and a subtraction module M107. A residual signal generation module M108, a transform module M109, a quantization module M110, an inverse quantization module M111, an inverse transform module M112, an addition module M113, and an entropy encoding module M114.

  In one embodiment, the block division module M101, the sub-partition generator module M102, the storage module M103, the motion detection module M104, the prediction signal generation module M105, the motion prediction module M106, the subtraction module M107, the residual signal generation module M108, the transformation The module M109, the quantization module M110, the inverse quantization module M111, the inverse transformation module M112, the addition module M113, and the entropy encoding module M114 are the block divider 501, the small partition generator 502, the frame memory of the moving picture encoding apparatus 10. 503, motion detector 504, prediction signal generator 505, motion predictor 506, subtractor 507, residual signal generator 508, converter 509, quantizer 510, inverse quantizer 511, inverse transformer 512, Adder 513, the same function as the entropy encoder 514, respectively causing a computer to execute C10. According to the moving picture coding program P10, the computer C10 can operate as the moving picture coding apparatus 10.

  As shown in FIG. 39, the moving picture decoding program P20 includes a data decoding module M201, a motion prediction module M202, an addition module M203, an inverse quantization module M204, an inverse transformation module M205, a storage module M206, a prediction signal generation module M207, and , And an addition module M208.

  In one embodiment, the data decoding module M201, the motion prediction module M202, the addition module M203, the inverse quantization module M204, the inverse transformation module M205, the storage module M206, the prediction signal generation module M207, and the addition module M208 are the moving picture decoding device. The same functions as the 20 data decoder 601, motion predictor 602, adder 603, inverse quantizer 604, inverse transformer 605, frame memory 606, predicted signal generator 607, and adder 608 are added to the computer C10. Let it run. According to the moving picture decoding program P20, the computer C10 can operate as the moving picture decoding apparatus 20.

  Various embodiments have been described in detail above. However, the present invention is not limited to the above embodiment. The present invention can be variously modified without departing from the gist thereof.

  DESCRIPTION OF SYMBOLS 100 ... Image predictive coding apparatus, 101 ... Input terminal, 102 ... Block divider, 103 ... Prediction signal generator, 104 ... Frame memory, 105 ... Subtractor, 106 ... Converter, 107 ... Quantizer, 108 ... Inverse Quantizer, 109 ... Inverse transformer, 110 ... Adder, 111 ... Quantized transform coefficient encoder, 112 ... Output terminal, 113 ... Predictive block division type selector, 114 ... Motion information estimator, 115 ... Prediction information Memory 116, Prediction information encoder 201 201 Input terminal 202 Data analyzer 203 Inverse quantizer 204 Inverse transformer 205 Adder 206 Output terminal 207 Quantization transform Coefficient decoder, 208 ... Prediction information decoder, 10 ... Video encoding device, 20 ... Video decoding device, 501 ... Block divider, 502 ... Subdivision generator, 503 ... Frame memory 504 ... Motion detector, 505 ... Prediction signal generator, 506 ... Motion predictor, 507 ... Subtractor, 508 ... Residual signal generator, 509 ... Converter, 510 ... Quantizer, 511 ... Inverse quantizer, DESCRIPTION OF SYMBOLS 512 ... Inverse converter, 513 ... Adder, 514 ... Entropy encoder, 601 ... Data decoder, 602 ... Motion predictor, 603 ... Adder, 604 ... Inverse quantizer, 605 ... Inverse converter, 606 ... Frame memory, 607 ... Prediction signal generator, 608 ... Adder, 5061 ... Motion vector memory, 5062 ... Motion reference destination candidate generator, 5063 ... Prediction motion vector generator, 6021 ... Motion vector memory, 6022 ... Motion reference destination candidate Generator, 6023 ... Predictive motion vector generator.

Claims (1)

Out of compressed image data encoded by dividing into a plurality of regions, encoded data of prediction information that indicates a prediction method used for predicting a signal of a target region to be decoded, and a prediction signal of the target region A data analysis step of extracting the encoded data and the encoded data of the residual signal;
The encoded data of the prediction information is decoded, and the prediction block division type indicating the number and the shape of a plurality of prediction areas, which are subdivision areas of the target area, and the prediction signal of each prediction area are obtained from the already reproduced signal. Predictive information decoding step for restoring motion information to
A prediction signal generating step for generating a prediction signal of the target region based on prediction information associated with the target region;
A residual signal restoration step of restoring the reproduction residual signal of the target area from the encoded data of the residual signal;
A recording step of restoring the pixel signal of the target area by adding the prediction signal and the reproduction residual signal, and storing the pixel signal as the already reproduced signal;
Including
In the prediction information decoding step,
Decoding the prediction block division type of the target area, and storing the prediction block division type in a prediction information storage unit that stores the decoded prediction information;
Generation of a prediction signal of a target prediction region that is a next prediction region based on prediction information of an adjacent region adjacent to the target region, the number of decoded prediction regions in the target region, and decoded prediction information of the target region Selecting candidate motion information to be used from the decoded motion information of the region adjacent to the target prediction region,
In accordance with the number of selected motion information candidates, merge block information and motion information for instructing generation of a prediction signal of the target prediction region using the selected motion information candidates, or the merge block information and the motion information Any one of the above, and the motion information used to generate the prediction signal of the target prediction region is stored in the prediction information storage unit,
An image predictive decoding method characterized by the above.

Images